Documente Academic
Documente Profesional
Documente Cultură
Hanno Essén
Royal Institute of Technology,
Department of Mechanics
Stockholm, Sweden
i
ii CONTENTS
6 Impact 93
6.1 The Impact Phenomenon . . . . . . . . . . . . . . . . . . . . . . . . . . 93
6.2 Impact and the Momentum Principles . . . . . . . . . . . . . . . . . . . 94
6.3 The Coefficient of Restitution . . . . . . . . . . . . . . . . . . . . . . . . 97
6.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
6.5 Hints and Answers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
iii
iv LIST OF FIGURES
This chapter presents the principles of linear and angular momentum in the form they
take for a system of particles. The elimination of the internal forces and moments is
discussed. The forms that the principle of angular momentum takes for some special
cases are derived and applied.
If we multiply and divide with this in the left hand side of the previous equation we
get, using the centre of mass definition
N
j=1 mj rj
rG = N , (1.5)
j=1 mj
1
2 CHAPTER 1. ANGULAR MOMENTUM OF PARTICLE SYSTEMS
the relationship N
j=1 mj vj
p = m N = mvG . (1.6)
j=1 mj
So, for a system of particles we find that the momentum can be expressed as total mass
times velocity of centre of mass:
p = mvG . (1.7)
The angular momentum of a system is also defined as the vector sum of the the
corresponding quantities for the individual particles
N
N
LA = LAj = AP j × mj vj . (1.8)
j=1 j=1
Here we can again attempt to introduce the centre of mass but things will not become
as simple as for linear momentum. We first put AP j = AG + GP j and, since A is
assumed to be at rest, we get
d d d
vj = AP j = (AG + GP j ) = (AG + rj ) = vG + vj . (1.9)
dt dt dt
Here rj = GP j is the position vector of particle j with respect to the centre of mass
and vj is the velocity of particle j with respect to the centre of mass or, equivalently,
the velocity of particle j in the ‘centre of mass system’. According to their definition
these quantities obey
N
N
mj rj = mj vj = 0, (1.10)
j=1 j=1
see figure 1.1. We now rewrite the angular momentum using this:
N
LA = (AG + rj ) × mj (vG + vj ). (1.11)
j=1
When the parentheses are expanded we get four terms but two of these are seen to give
the two parts of zero because of the relationships 1.10 so finally
the angular
momentum
N
LA = AG × mvG + rj × mj vj
j=1
⇐⇒
LA = AG × p + LG . (1.12)
The total angular momentum of a system is thus the sum of two parts: one, external,
which only refers to the centre of mass of the system and one, internal, LG , which
depends only on the positions and velocities of the particles with respect to the centre
of mass system. When the system reduces to a single particle this second internal part
vanishes.
The angular momentum with respect to the centre of mass, LG , is by definition
N
LG = GP j × mj vj (1.13)
j=1
N
N
LG = rj × mj (vG + vj ) =( mj rj ) × vG + LG . (1.14)
j=1 j=1
1.1. LINEAR AND ANGULAR MOMENTUM OF PARTICLE SYSTEMS 3
P Z'
2
P r'3
3 r2 r'
2
r3 G Y'
Z
OG r'1
X'
P
1
r1
O Y
Figure 1.1: This figure shows the definition of the centre of mass system (G, X , Y , Z )
of a three particle system. The position vectors of the particles in the fixed system,
OP i = ri (i = 1, 2, 3), are shown with filled black heads. The position vectors of the
particles in the centre of mass system, GP i = ri , are shown with dashed heads. By
definition we then have that mi ri = mOG while mi ri = mGG = 0.
4 CHAPTER 1. ANGULAR MOMENTUM OF PARTICLE SYSTEMS
f12 P P
1 f12 1
r21 r21
P2
f 21
P
f 21 2
Figure 1.2: Two examples of internal (inter particle) forces that both obey Newton’s third law
f12 = −f21 . In the example on the left the forces are parallel to the vector between the particles,
f21 r21 . In the example on the right this is not the case. Such cases must be excluded if the
angular momentum principle for systems is to hold in its usual form.
where the superscript ‘i’ stands for internal and ‘e’ stands for external. The internal
force on a given particle can always be thought of as arising from the other particles
of the system and it can therefore be expressed as the vector sum of of contributions
from the rest of the system. If we denote the force from particle j on particle k by
It is natural to define fkk = 0 so that the term for j = k does not have to be excluded
from the sum. This is also a natural consequence of Newton’s third law,
d N N
ṗ = mj vj = mj r̈j = (1.20)
dt j=1 j=1
1.1. LINEAR AND ANGULAR MOMENTUM OF PARTICLE SYSTEMS 5
N
N
= Fj = (Fij + Fej ) = (1.21)
j=1 j=1
N
N
= fjk + Fe = Fi + Fe . (1.22)
j=1 k=1
Here Fe denotes the force sum of the external forces. The double sum over the internal
forces gives 0 because of the relations 1.19 so that
Fi = 0, (1.23)
d N N
L̇A = AP j × mj vj = (vj × mj vj + AP j × mj r̈j ) = (1.25)
dt j=1 j=1
N
N
N
= AP j × (Fij + Fej ) = AP j × Fij + AP j × Fej = (1.26)
j=1 j=1 j=1
N
= AP j × Fij + MeA = MiA + MeA . (1.27)
j=1
Here MeA stands for the moment of the external forces with respect to A. The sum of
the moments of the internal forces corresponds to the sum that became zero when we
derived the momentum principle. We can rewrite it as follows:
N
N
N
MiA = AP j × Fij = (AP j × fjk ) = AP j × fjk = (1.28)
j=1 j=1 k=1 k,j
1
= AP j × fjk + AP k × fkj = (1.29)
2 k,j k,j
1 1
= AP j × fjk + AP k × fkj = AP j − AP k × fjk = (1.30)
2 k,j 2 k,j
1 1
= P k P j × fjk = rjk × fjk = Mi , (1.31)
2 k,j 2 k,j
where rjk ≡ rj − rk . So Newton’s third law makes the internal moment independent of
base point, but it does not make it zero. This expression shows that the sum will become
zero provided that the force from particle k to particle j is parallel to the vector from
particle k to particle j, i.e. if fjk rjk . This behavior of inter-particle forces agree with
those of the gravitational and electrostatic interactions. When magnetic interactions
are taken into account, however, it may be violated. Since magnetic forces are many
orders of magnitude weaker than the corresponding Coulomb forces between charges
it seems as if it might at least be a good approximation to neglect this sum. We thus
assume that
MiA = 0, (1.32)
and obtain the principle of angular momentum for a system, on the form
6 CHAPTER 1. ANGULAR MOMENTUM OF PARTICLE SYSTEMS
Z
v
m
ϑ Y
Figure 1.3: This figure shows the skew rotating two particle system discussed in example 1.1.
the principle of
angular
momentum L̇A = MeA . (1.33)
That is, the time derivative of the angular momentum vector is equal to the total
moment of the external forces. This equation, or something equivalent which amounts
to the fact that the internal moments sum to zero, is often postulated in classical
mechanics. There are magnetic phenomena in physics where it seems to be violated
but this only means that the electro-magnetic field can carry an angular momentum
which is not taken into account by classical mechanics. In conclusion, one can safely
use equation 1.33, the angular momentum principle, when solving problems in classical
mechanics.
Example 1.1 At each end of a light rod of length 2R there is a small, heavy ball of mass
m. At its midpoint the rod is fixed to a rotating axis with which it makes an angle ϑ. The
axis rotates, with fixed direction, with constant angular velocity ω. Calculate the momentum,
angular momentum, external force and moment, with the midpoint of the rod as base point,
for this two particle system.
Solution: We choose the coordinate system as in figure 1.3, so that the rotation axis is along
the Z-axis. For the particle with positive z-coordinate we then have
while the other particle has r2 = −r1 and v2 = −v1 . This means that the linear momentum is
LO = r1 × mv1 + r2 × mv2 = 2mR2 sin ϑ ω[− cos ϑ(cos ωt ex + sin ωt ey ) + sin ϑ ez ] (1.37)
Note that LOz = 2mR2 sin2 ϑ ω =const. but that LO (t) as a whole is not. By taking the time
derivative we find
The rotation axis must thus act with this moment on the light rod at its midpoint. ✷
1.2. MOVING THE BASE POINT 7
1.1.3 Summary
To sum up this section we have found that the two principles of linear and angular
momentum for a particle also hold, in the same form, for systems of particles provided
the total force and moment acting on the particle are replaced by the sum of the external
forces and their moments, respectively:
ṗ = Fe , (1.39)
L̇A = MeA . (1.40)
The superscript ‘e’ on the right hand side quantities appear because we have needed to
carefully distinguish between internal and external forces in this section. In what follows
we will assume that it has been understood that only the external forces contribute
and skip the superscript. The momentum and angular momentum appearing here can
be expressed as follows
p = mvG , (1.41)
LA = AG × p + LG . (1.42)
The centre of mass motion of a particle system or body is thus just like the motion of
a particle, but the angular momentum will behave differently because of the presence
of the internal part LG . We will investigate this further below.
N
N
LB = LBj = BP j × mj vj = (1.43)
j=1 j=1
N
N
N
= (BA + AP j ) × mj vj = BA × mj vj + AP j × mj vj (1.44)
j=1 j=1 j=1
= BA × p + LA . (1.45)
We have thus derived the connection formula for angular momentum connection
formula for
LB = LA + BA × p. (1.46) angular
momentum
This formula is of the same form and is derived in the same way as the corresponding
formula for the moment of a force system: MB = MA + BA × F.
We now put this expression into L̇A = MA and use the fact that ṗ = F:
L̇B + vB × p + AB × F = MA . (1.49)
When the last term on the left hand side is moved to the right and the connection
angular formula for moments is taken into account we end up with the result
momentum
principle for L̇B + vB × p = MB . (1.50)
moving base
point The angular momentum principle must thus have the additional term vB × p on the
left hand side when the base point B moves with velocity vB . When this velocity is
zero this formula correctly reduces to the old result.
The use of a moving base point is particularly convenient when the extra term
vanishes. There are two (non-trivial) cases when this happens. Firstly is clearly zero if
p = mvG = 0, i.e. if the centre of mass of the body is at rest. Secondly it is zero if the
the two vectors of the vector product are parallel, that is, if vG vB . This happens if
the base point moves so that its velocity is parallel to that of the centre of mass.
L̇G = MG (1.52)
which means that one may calculate the angular momentum using quantities in the
centre of mass system.
With the help of the connection formula for moments, MB = MA + BA × F, and
ṗ = F, one can rewrite the above formulae so that one gets, for example,
L̇G + AG × ṗ = MA . (1.53)
The time integral of the external force is a vector quantity, the impulse of the force and
we write t2
I≡ F dt. (1.55)
t1
If a force acts continuously the impulse will depend on the two integration limits.
Mostly, however, one speaks of the impulse of a force that is non-zero only for some
finite time. The quantity I is then independent of the integration interval as long
1.4. CONTINUOUS MASS DISTRIBUTIONS 9
it contains the time-interval when the force is non-zero. The integrated form of the
momentum principle can also be written
∆p = I (1.56)
which says that the change in momentum for a system, or a body, is given by the
impulse of the ( external) force on the system. If there is no external force there will
consequently be no change in the momentum, i.e.
one finds that the change in the angular momentum vector ∆LA over some time interval,
is given by the angular impulse delivered by the external moment during this time
interval
∆LA = HA . (1.60)
Should the (external) moment on the system be zero the angular momentum will be
constant and we have the conservation law for angular momentum:
The conservation laws for momentum and angular momentum are often useful in prob-
lem solving. Note that these equations are vector equations and that they thus really
correspond to three real equations each. With suitable choice of basis vector directions
one can often find that the conservation laws apply, if not in full, at least for one or
more of the component equations.
The ideas of impulse and angular impulse are of particular importance when large
forces act during short times, i.e. when one has impact. We will return to this later.
Many of the derivations that we have made above become more intricate when we
must consider integrals over the bodies but these difficulties are mainly of a technical
nature and it is often taken as a postulate that the same laws hold for for continuously
distributed matter as for systems of particles. We shall adhere to this view in this text.
We will, however, mostly present the general definitions and derivations assuming the
particle system description and pass to the integrals over continuous distributions only
when it is convenient.
Recall that if the point A is on the Z-axis then MAz = Mz is the moment with respect
to the Z-axis.
Let us now calculate the z-component of the angular momentum. To simplify the
notation we put the origin at the base point A = O. We now have
N
LOz = (OP j × mj vj ) · ez (1.66)
j=1
N
N
= mj (rj × vj ) · ez = mj (xj ẏj − yj ẋj ). (1.67)
j=1 j=1
so that
N
LOz (t) = mj ρ2j (t)ϕ̇j (t). (1.69)
j=1
If we now define the average angular velocity of the body (with respect to the Z-axis)
by
N 2
j=1 mj ρj (t)ϕ̇j (t)
ωav (t) ≡ N 2
(1.70)
j=1 mj ρj (t)
and the moment of inertia, Jz , of the body with respect to the Z-axis by
N
Jz ≡ mj ρ2j (1.71)
j=1
ρ
. Y
2 m 1
m2
v2 ϕ
ρ
1 v1
.
ρϕ ρ
2
2 2 2 ϕ
1
X
ϕ
3 ρ
3
v3
m 3
Figure 1.4: A system of three particles viewed along the Z-axis, which points up from the
paper. The cylindrical coordinates ρi , ϕi (i = 1, 2, 3) of the particles are shown as well as their
velocity vectors vi . For particle number 2 the components of the velocity along the radial (eρ )
direction and the eϕ -direction are indicated.
We can now apply these results to the z-component of the angular momentum
principle, equation 1.65, in the form
We find that
J˙z ωav + Jz ω̇av = MOz , (1.74)
and this is thus a general expression for the z-component of the angular momentum
principle.
If all particles have the same angular velocity ϕ̇ then, of course, ωav = ϕ̇. The most
important case for which this happens is when the system is rigid and rotates around
the Z-axis. In general, however, it will happen whenever the velocities of the particles
are such that the ϕ̇i all are equal while ρ̇i and żi are arbitrary. For these cases, when
the angular velocity is well defined, one can simply write
If we furthermore assume that all ρi are constant, which they will be if the body is rigid
and rotates around the Z-axis, then Jz is constant. In this case equation 1.74 reduces
to
Jz ϕ̈ = MOz , (1.76)
i.e. the angular accelaration is simply proportional to the moment.
If there is no external moment with respect to the Z-axis through O so that MOz = 0
then the z-component of the angular momentum vector will be conserved, LOz =const.,
and this thus implies that
Jz ωav = const. (1.77)
This equation tells us that a large moment of inertia Jz implies small (average) angular
velocity and vice versa, a fact used by springboard divers and figure skaters, see figure
1.5.
Example 1.2 A person walks on a horizontal platform that can rotate freely around a vertical
Z-axis. The empty platform has the moment of inertia Jz . The person P, which can be treated
12 CHAPTER 1. ANGULAR MOMENTUM OF PARTICLE SYSTEMS
Figure 1.5: This picture shows a person on a platform that can rotate with negligible friction
around a fixed vertical axis. Then Jz (t)ωav (t) = constant and this means that the angular
velocity is larger when the weights in the hands are held close to the body and the body is close
to the rotation axis, and vice versa.
Y'
B
ρ P
ϕ
O
X
A
X'
−α
Figure 1.6: This figure shows the platform discussed in example 1.2 as seen from above. It
also shows the path that the person walking on the platform has taken between points A and
B of the platform and the cylindrical coordinates (ρ, ϕ) of a point P on this path relative to
a coordinate system fixed to the platform. The angle −α is the angle that the platform has
rotated relative to a fixed direction (the X-axis) when the person has walked from the initial
point A to P.
1.5. PROJECTION ON A FIXED DIRECTION 13
as a particle, has mass m. The position of the person relative to a coordinate system OX Y Z
fixed to the platform is measured in terms of cylindrical coordinates ρ and ϕ, see figure 1.6.
a) Find a relationship between the small change in angle dϕ of the person and the corresponding
absolute rotation angle dα of the platform relative to a fixed direction (the X-axis).
b) Use this relationship to calculate the rotation angle ∆α of the platform when the person
walks along a given path (ρ(t), ϕ(t)) on the platform from points A, with ϕ = 0, to B, with
ϕ = ∆ϕ, and show that this angle depends in general on the path.
Solution:
a) There is no external moment with respect to the Z-axis on the system of particles defined
by the platform and the person on it. Therefore Lz = constant. There are two contributions
to Lz , that of the platform, which is Jz α̇, and that of the person which is given by mρ2 (α̇ + ϕ̇).
Note that the angular velocity of the person with respect to the fixed system is equal to the
angular velocity of the platform, α̇, plus the angular velocity of the person with respect to the
platform, which is ϕ̇. We thus have
If we have initial conditions such that α̇(0) = ϕ̇(0) = 0 we get Lz = 0 and the above equation
can be written
dα dα dϕ
0 = Jz + mρ2 ( + ). (1.79)
dt dt dt
This gives us the relation
dϕ
dα mρ2
=− dt (1.80)
dt Jz + mρ2
between the two angular velocities.If we multiply by dt we get the differential relation
mρ2 dϕ
dα = − (1.81)
Jz + mρ2
for small angles.
b) If we integrate the relation 1.80 between the time t = 0, when the person is at A, and
the time t = T , when the person is at B, we get
T T ∆ϕ
mρ2 (t)ϕ̇(t) mρ2 (ϕ)dϕ
∆α = α̇ dt = − dt = − . (1.82)
0 0 Jz + mρ2 (t) 0 Jz + mρ2 (ϕ)
This integral will obviously depend on e.g. the function ρ(ϕ). If the radius ρ is kept constant,
ρ(ϕ) = ρ0 , during the walk, the integral can be evaluated and one finds the relation
mρ20
∆α = − ∆ϕ (1.83)
Jz + mρ20
between the two angles. Note that the angle of rotation of the platform is of opposite sign to
that of ∆ϕ as indicated in figure 1.6. ✷
14 CHAPTER 1. ANGULAR MOMENTUM OF PARTICLE SYSTEMS
Z
Y
ω
a b
Figure 1.7: This picture refers to problem 1.3 and shows the rod, its rotation axis, and the
placement in the coordinate system.
1.6 Problems
Problem 1.1 A particle P of mass m has constant speed v in a circular trajectory of
radius r in the xy-plane with centre at the origin of the coordinate system.
a) Calculate the angular momentum LA with respect to a point A with Cartesian
coordinates (a, b, c).
b) Then calculate the time derivative of this quantity to get MA and investigate whether
the point A can be chosen so that MA = 0.
c) Use the result of b) to find the direction of the force acting on the particle.
Problem 1.2 A three particle system consists of particles with masses 3m, m, and
5m. Their position vectors have Cartesian components (t, −2, 3t2 ), (t − 1, t3 , 5), and
(2 − t2 , t, t3 ) respectively. Calculate as functions of time
a) the total force acting on the system,
b) the total moment with respect to the origin.
Problem 1.4 In order to measure the moment of inertia JL of the complicated rotor
of an electric motor one mounts the rotor on bearings of negligible friction so that it can
rotate freely around a horizontal axis L. A thin flexible string is then wound around
the axis of the rotor. This axis has radius r. A mass M is then hung in the string and
it is found that, when starting from rest, a length x of the string becomes unwound in
time T . What is JL ? (See figure 1.13 in the hints and answer section if necessary.)
1.6. PROBLEMS 15
B ω
b
a
A
Figure 1.8: This picture refers to problem 1.6 and shows the disc, the lever arm AB, the
counterweight at B. It also indicates the vertical rotation axis of the disk and the fixed vertical
rotation axis of the horizontal lever arm.
Problem 1.5 A person stands at rest on a horizontal platform that can rotate freely
around a fixed vertical axis. By turning the upper body relative to the lower the
person tries to look backwards. When the system is in this position the upper body
has been turned an angle α while the lower part of the body together with the platform
has turned an angle β, both with respect to the same fixed direction. The moment
of inertia of the upper body is J1 while that of the lower body plus platform is J2 .
Calculate α as a function of the relative ‘twist’ angle α − β.
Problem 1.6 A homogeneous circular disc of mass M and radius R is being rotated
by a small electric motor at A, around a vertical axis, see figure 1.8. The electric motor
has mass m and is mounted at one end of a horizontal lever arm AB. At the other end
B of the lever arm, which can rotate freely around a fixed vertical axis at the distance
a from A, there is a counterweight. The distance of the counterweight from the fixed
axis is b and it balances the mass at the other end of the arm. The power supply of
the motor is suddenly cut off and the disc starts to slow down due induction effects in
the electric motor. Assume that the angular velocity of the disc initially was ω while
the lever arm was at rest. Calculate the final angular velocity of the disc.
Problem 1.7 A homogeneous solid sphere of mass M and radius R is rotating freely
with angular velocity ω0 around a vertical axis through its centre of mass. Along a
horizontal diameter a smooth narrow channel has been drilled through the sphere. Two
small balls each of mass m are initially at rest in the middle of the channel, see figure
1.9. A small charge between the balls suddenly explodes and gives them initial speeds
v in opposite directions. When the balls leave the sphere the angular velocity has
changed. Calculate the change in angular velocity and explain how it came about.
R
m m R/2
ω 0 R
Figure 1.9: The figure on the left refers to problem 1.7. Two small balls are initially in a
drilled channel in the middle of the rotating sphere.
Figure 1.10: The figure on the right refers to problem 1.8 and shows, from above, the closed
path followed by the person on the freely rotating platform.
Figure 1.11: The figure on the left refers to problem 1.10. Two particles are attached to two
springs symmetrically placed in a smooth pipe which can move on a smooth horizontal plane.
Figure 1.12: The figure on the right refers to problem 1.11. The cylinder starts with the
velocity and the angular velocity shown in the figure. It moves on a rough horizontal plane.
is held to the string at B so that it burns off. Calculate the ratio of the tensions in the
other string before and just after the burning.
Problem 1.10 A homogeneous straight smooth pipe of mass M and length a can
move on a smooth horizontal plane. In the middle of the pipe the ends of two identical
springs, each of unloaded length L and stiffness k, are attached. At the other ends of
the springs there are particles of mass m attached, see figure 1.11. At time t = 0 the
two particles are pulled out a distance b from their symmetric equilibrium positions
inside the pipe and are then released from rest. The pipe is at the same time given an
angular velocity ω0 . Describe the qualitative motion of the pipe.
Problem 1.11 A homogeneous circular cylinder of mass m and radius r is given a
speed v and an angular velocity ω on a rough horizontal floor with coefficient of (kinetic)
friction f . Determine the value of ω so that the cylinder returns to the starting point
after a tour along the floor. See figure 1.12.
1.7. HINTS AND ANSWERS 17
Mg
Figure 1.13: This picture refers to answer 1.4 and is a view of the rotor along its axis (of
radius r). It also shows the string, wound round the axis, in which the mass M hangs down.
The forces acting on M are indicated.
Answer 1.2
a) F = mk r̈k = 2m(−5, 3t, 9 + 15t),
b) MO = rk × mk r̈k = 2m(−18 − 15t + 15t2 , −39t + 10t3 , 2t + 3t2 ).
Answer 1.3
a) Elements of the rod of length dx have mass dm = m dx/(a + b) and the velocities of
these elements are
v(x) = ẏ(x) = (x − c)ω
so that they are negative for x < c and positive otherwise. Now
2 a2 + b2 − ab
c=
3 a−b
which thus is the answer. One notes that when a → b it is not possible to make Lz = 0
except in the limit c → ∞.
Answer 1.4
The equations of motion can be written
JL ϕ̈ = rS,
M ẍ = M g − S.
18 CHAPTER 1. ANGULAR MOMENTUM OF PARTICLE SYSTEMS
See figure 1.13 for notation. The length of unwound string x must obey
x = rϕ ⇒ ẍ = rϕ̈.
This system of equations is easily solved and then integrated with respect to time.
Finally solving for JL gives
1 M gr2 2
JL = T − M r2
2 x
for the moment of inertia of the rotor.
Answer 1.5
The answer is α = J2 (α − β)/(J1 + J2 ). Note the limits J2 → 0 and J2 → ∞.
Answer 1.6
Use the fact that Lz is conserved since there is no external moment with respect to
the fixed vertical axis through the lever arm. Also note that the disc will slow down
until it has the same angular velocity as the motor and lever arm. The mass m of the
counterweight is calculated from the relation b m = a(M +m). The relation J0 ω = J1 ω
with J0 = 12 M R2 and J1 = (J0 + M a2 ) + ma2 + m b2 then gives
ω
ω =
1+ 2 (MM
+m) a(a+b)
R2
Answer 1.7
Use the fact that Lz is conserved for the system as a whole. The two balls contribute
the amount 2mr[rω(r)] to Lz when at radius r. Here ω(r) is the angular velocity of
the sphere when the ball are at radius r. The initial angular velocity is thus, with this
notation, ω0 = ω(0). The moment of inertia of the sphere itself is Js = 25 M R2 , the
balls contribute the amount Jb = 2mr2 , when at radius r. The final angular velocity
is thus ω(R) = ω(0)M/(M + 5m) and the this gives
5m
∆ω = ω(0) − ω(R) = ω0
M + 5m
for the decrease.
Answer 1.8
The angle is given by
π/2 0 m
M R2 dϕ M (R/2)2 dϕ π 3M
∆α = − 1 2 2
− 1 2 2
=− m m .
0 2 mR + M R π/2 2 mR + M (R/2)
2 (2 + M )(1 + 2 M )
Note that it goes to zero when m/M → ∞ i.e. when the platform becomes heavy. It
also becomes zero when the platform has negligible mass (m = 0). If the numerator
and denominator of the formula are both multiplied by M 2 /m2 the simplified result is
the same formula with m/M replaced by M/m.
Answer 1.9
Before the burning the tension in each string must be S = mg/2. Immediately after
the burning the bar is still horizontal. With a vertical Y -axis and with ϕ the angle
that the bar makes with the horizontal the principles of linear and angular momentum
give the equations of motion:
mÿG = S − mg,
JG ϕ̈ = S#/2,
1.7. HINTS AND ANSWERS 19
# # #
yG = − sin ϕ ⇒ ẏG = − ϕ̇ cos ϕ ⇒ ÿG = (ϕ̇2 sin ϕ − ϕ̈ cos ϕ)
2 2 2
The initial conditions ϕ(0) = ϕ̇(0) = 0 thus give
#
ÿG = − ϕ̈.
2
This equation together with the two equations of motion give us a system of three
equations for three unknowns (S, ÿG , ϕ̈). Solution of this system of equations gives us
the result S = mg/4. The ratio of the two tensions is thus
S /S = (mg/2)/(mg/4) = 2
Answer 1.10
Since there are no external forces on the pipe its centre of mass will move with constant
velocity; if this velocity is initially zero the centre of mass (the middle of the pipe) will
remain at rest. The two particles will oscillate radially in the pipe and this means that
the moment of inertia of the pipe will vary and therefore the angular velocity of the
pipe will also vary.
Answer 1.11
The velocity of the centre of mass of the cylinder must have changed direction when
rolling without slipping occurs. This gives the condition ω > 2v/r.
20 CHAPTER 1. ANGULAR MOMENTUM OF PARTICLE SYSTEMS
Chapter 2
This chapter presents rigid body kinematics. This means that methods for describing
the position, the orientation, and the velocity state of the rigid body are presented.
21
22 CHAPTER 2. KINEMATICS OF RIGID BODIES
φ
d
Figure 2.1: This figure illustrates translation of a rigid body (on the left). A translation is a
displacement of the body such that all points of the body undergo the same translation. On
the right a rotation is illustrated. A rotation of a rigid body is a displacement of the body such
that one point, of the body or rigidly connected to the body, remains fixed.
but one must keep in mind that finite rotations do not add vectorially. The symbol
φ is, in this sense, only a convenient way of writing the three parameters specifying
the rotation. We will see later, however, that infinitesimal rotations do add, so that it
is meaningful to write for example δ φ = δ φ1 + δ φ2 . Rotations around a fixed axis
direction also are additive in this way.
To summarize we see that we need six numbers, three coordinates and three angles,
to specify the position and orientation of a rigid body. We can take the reference state
of the body to be one for which the point A coincides with the origin and for which
it has some given orientation. To bring it to any other position one then translates
it with d = rA without rotation. One then rotates it around an axis through A with
rotation vector φ to bring it to the desired orientation. One says that the rigid body
has six degrees of freedom and the six coordinates can be taken as the six components
of the vectors rA , φ.
The translational degrees of freedom of the body are similar to those of a particle
which we already know about. The rotational degrees of freedom are, however, quite
different and in order to understand the kinematics of the rigid body we must now
study the properties of rotations in greater detail.
Figure 2.2: In example 2.1 the rotation matrix connecting the two bases in this figure according
to formula 2.10 is given.
basis vectors
eO
1 eA
1
EO = eO
2 and EA = eA
2 , (2.8)
eO
3 eA
3
of direction cosines of the body-fixed basis vectors with respect to the basis vectors
fixed in the observer reference frame. The equation connecting the two bases can now
be written on the form
EA = ARO EO (2.10)
Example 2.1 Calculate the rotation matrix 2.9 explicitly for the case when EA is obtained
by rotating the basis EO the angle ψ around eO 3 in the positive sense, see figure 2.2.
Solution: The figure shows that eA 1 +sin ψ e2 and that e2 = − sin ψ e1 +cos ψ e2 .
O O A O O
1 = cos ψ e
This means that the rotation matrix becomes
cos ψ sin ψ 0
R = R3 (ψ) = − sin ψ cos ψ 0
A O
(2.11)
0 0 1
and, as indicated, we will denote this matrix by R3 (ψ) which thus stands for a matrix that
rotates an angle ψ around the third basis vector. ✷
The rotation matrix ARO , being a 3 × 3 matrix, has nine elements but these cannot
all be independent since the rows of the matrix are the components of orthonormal
24 CHAPTER 2. KINEMATICS OF RIGID BODIES
basis vectors in an orthonormal basis. Like any orthonormal basis the body-fixed basis
is characterized by the six relations
1 if i = j,
i · ej = δij ≡
eA A
(2.12)
0 if i = j.
between the scalar products. When these are written in terms of the components they
give the six conditions
cos αi · cos αi + cos βi · cos βi + cos γi · cos γi = 1 for i = 1, 2, 3
(2.13)
cos αi · cos αj + cos βi · cos βj + cos γi · cos γj = 0 for i, j = 1, 2; 1, 3; 2, 3
which the nine elements of the of the rotation matrix 2.9 must obey. The matrix can
thus depend only on (9 − 6 = 3) three independent parameters. A rotation is thus
completely specified by three parameters and this is in agreement with the statement
of the previous section where we explained how a rotation can be specified by three
components of the rotation vector φ.
The matrix equation 2.10 can be ‘solved’ for the basis EO in terms of EA by multi-
plying both sides with the inverse of ARO . We get
(ARO )−1 EA = (ARO )−1 ARO EO = 1 EO = EO . (2.14)
In order for our notation to be consistent we should have EO = ORA EA so that
(ARO )−1 = ORA . But if we interchange A and O in the first matrix of formula 2.9,
to get ORA , then we get a new matrix with rows which are equal to the columns of the
old and vice versa. A matrix obtained by interchanging rows and columns in another
matrix is called the transpose of the old and is denoted by a superscript T . We have
now shown that
(ARO )−1 = (ARO )T . (2.15)
A matrix which has this property, that the inverse is equal to the transpose, is called
an orthogonal matrix and the rotation matrices are thus orthogonal matrices. This
fact is closely related to the fact that the elements in the rows of such matrices are the
components of orthonormal basis vectors in an orthonormal basis. This is best seen by
the fact that the relations 2.13 (using 2.9) are equivalent to the matrix equations
A O O A
R R = ARO (ARO )T = ARO (ARO )−1 = 1
⇐⇒ (2.16)
cos α1 cos β1 cos γ1 cos α1 cos α2 cos α3 1 0 0
cos α2 cos β2 cos γ2 cos β1 cos β2 cos β3 = 0 1 0 .
cos α3 cos β3 cos γ3 cos γ1 cos γ2 cos γ3 0 0 1
Figure 2.3: This figure illustrates the effect of the rotation implied by the rotation vector φ on
an arbitrary vector r. Note that |ea × er | = sin α so that the three vectors with open triangular
heads make up an orthonormal triad of basis vectors. When the rotated vector r is expressed
in terms of these one obtains formula 2.23.
If more than one basis is involved in the problem at hand we will indicate this with
superscripts on the relevant matrices, components, etc, so that we write
eO
1
O
r = rO EO = (xO O O A A
1 x2 x3 ) e2 = r E . (2.19)
O
e3
Note that the vector r itself is independent of the basis so that r = rO EO = rA EA . Use
of this and 2.10 and 2.16 shows that the components of a vector transforms according
to
rA = rO ORA . (2.20)
This equation essentially says that when we rotate the basis we must make a compen-
sating ‘rotation’ of the components in order to keep the vector the same. In order to
study the rotation of rigid bodies we, however, need to know how to actually rotate
vectors. One way of doing this is to rotate the basis without the compensating rotation
of the components. The vector then simply follows the basis in its rotation. To every
rotation matrix ARO there therefore corresponds a rotation operator R̂O→A defined by
an angle α = [φ, r] with the axis. If α = 0 the vectors in the figure become undefined
but then r is parallel to ea so the rotation has no effect. It should be clear from the
figure that
r = (r · ea ) ea + (ea × r) × ea (2.22)
and that the rotated vector is given by (note that r = r er )
The rotation matrix ARO corresponding to the rotation operator R̂(φ) will now be
calculated.
The two equations 2.21 and 2.23 can be used to calculate the elements of the rotation
matrix 2.9 explicitly in terms of the quantities of the rotation vector φ = φ ea . In order
to do this we insert the expressions
r = rO EO = xO O O O O O
1 e1 + x2 e2 + x3 e3 , (2.24)
ea = cos α1 eO
1 + cos α2 eO
2 + cos α3 eO
3 (2.25)
into equation 2.23. Here αi are the angles between the observer fixed (non-rotated)
basis vectors and the direction of the rotation axis ea . When this has been done it is
only a matter of some algebra to rewrite this equation (2.23) on the form
3
3
r = xO O
i Rij ej , (2.26)
i=1 j=1
where Rij stand for algebraic expression in terms of the quantities of the rotation vector.
When we compare this with the equation r = rO ARO EO written explicitly as a sum
over the matrix elements
3
3
3
3
r = (rO )i (ARO )ij (EO )j = xO A O O
i ( R )ij ej (2.27)
i=1 j=1 i=1 j=1
we see that the algebraic expressions Rij are in fact the matrix elements of the rotation
matrix ARO . When this program is carried out one finds the result
A O
R (φ) = (2.28)
(1 − cos φ) cos2 α1 + cos φ (1 − cos φ) cos α1 cos α2 + sin φ cos α3 (1 − cos φ) cos α1 cos α3 − sin φ cos α2
(1 − cos φ) cos α2 cos α1 − sin φ cos α3 (1 − cos φ) cos2 α2 + cos φ (1 − cos φ) cos α2 cos α3 + sin φ cos α1
(1 − cos φ) cos α3 cos α1 + sin φ cos α2 (1 − cos φ) cos α3 cos α2 − sin φ cos α1 (1 − cos φ) cos2 α3 + cos φ
where the first and last matrices are seen to be symmetric and the middle one is
anti-symmetric. For small rotation angles φ the three terms in the sum represent
constant+quadratic, linear, and quadratic terms in φ, respectively. Note that the cos αi
are direction cosines of the axis direction so they obey cos2 α1 +cos2 α2 +cos2 α3 = 1 and
only two of them are independent. This means that this rotation matrix depends on
three independent parameters as we have discussed above. We will now find a different,
more direct way of parameterizing a rotation matrix.
2.4. EULER ANGLES AND NON-COMMUTATION OF ROTATIONS 27
Figure 2.4: This figure shows the effect of the first step of the three necessary to achieve
general rotation. The rotation is from the fixed basis triad EO to the intermediate basis EC
and corresponds to rotation an angle ψ around the direction given by eO
3.
This step is shown in figure 2.5. Note how the plane spanned by eB B
1 and e2 now is
B
tilted an angle θ. The orientation of the basis E is still not arbitrary since the basis
vector eB
1 necessarily is in the original 1,2-plane. This is now changed by the third step:
a rotation an angle ϕ around eB 3 . This is shown in figure 2.6 and the formula is
Any desired orientation of the basis triad EA can clearly be achieved with this sequence
of steps. If we now put it all together we have
Figure 2.5: This figure shows the effect of the second rotation from the basis triad EC to the
second intermediate basis EB . This rotation is an angle θ around eC
1 . The line of intersection,
parallel to eC B
1 = e1 , between the 1,2-plane of the O-system and the 1,2-plane of the B-system,
is sometimes called the line of nodes.
Figure 2.6: This figure shows the effect of the third and final rotation from the basis EB to
the new desired basis EA . The three angles, ψ, θ, and ϕ, which parameterize the total rotation
from EO to EA are called Euler angles. The symbols ψ̇, θ̇, and ϕ̇ indicate the axes around which
rotation takes place if the corresponding angle changes while the two others are held fixed.
2.5. INFINITESIMAL ROTATIONS AND ANGULAR VELOCITY 29
φ1
φ2
φ1
φ2
Figure 2.7: This figure illustrates the effect of two consecutive rotations of a rectangular box.
In the upper row the box is first rotated 90◦ around a vertical axis, then 90◦ around a horizontal
axis. In the lower row the same two rotations are performed in the opposite order. The end
results are clearly different and this shows that finite rotations do not commute; the order in
which they are done matters.
The parameters in the parameterization of the rotation matrix constructed in this way
are called Euler angles. The rotation matrix in terms of Euler angles ψ, θ, ϕ is thus
A O
R (ψ, θ, ϕ) = R3 (ϕ) R1 (θ) R3 (ψ). (2.35)
When the matrix multiplications are carried out explicitly one obtains
A O
R (ψ, θ, ϕ) = (2.36)
cos ϕ cos ψ − sin ϕ cos θ sin ψ cos ϕ sin ψ + sin ϕ cos θ cos ψ sin ϕ sin θ
− sin ϕ cos ψ − cos ϕ cos θ sin ψ − sin ϕ sin ψ + cos ϕ cos θ cos ψ cos ϕ sin θ ,
sin θ sin ψ − sin θ cos ψ cos θ
but this form is not particularly useful and is only given here for reference. By equating
the antisymmetric part of this matrix with the antisymmetric part in equation 2.29 one
can easily find the axis and angle of the net rotation produced by the three steps.
It is important to understand that the first and the last rotations R3 are not around
the same axis; the first is around eO B
3 but the last is around e3 . From this one realizes
that the order in which rotations are done is important; the end result will change if
the order is changed. As long as one rotates about a fixed axis this is not true since,
clearly,
R3 (ϕ) R3 (ψ) = R3 (ψ) R3 (ϕ) = R3 (ϕ + ψ). (2.37)
This, however, makes it obvious that
Algebraically this depends on the fact the matrix multiplication is not commutative but
the reader should also try to visualize the geometric difference between the two rotations
implied by the left and the right hand side. The non-commutation of rotations is also
illustrated in figure 2.7.
The effect of the antisymmetric matrix on the basis is thus equivalent to a vector
product. In the last term here the vector product in front of the column matrix EO of
basis vectors means that each vector in the matrix is to be multiplied vectorially by
δ φ = δφ1 eO O O
1 + δφ2 e2 + δφ3 e3 . The change in the basis vectors for a small rotation can
thus be written
δ EO = δ φ × EO (2.44)
by means of the vector product with the infinitesimal rotation vector.
We’ll now assume that the parameters of the rotation matrix are functions of time
so that the matrix describes the rotation of a rigid body A to which the basis EA is
attached with respect to the observer frame O. We wish to find the time derivatives of
the basis vectors of EA . The time derivative of a vector will depend on which reference
frame is considered as fixed. When it is the frame O which is fixed we denote this
O
by writing the time derivative dtd . We can now define a matrix which gives this time
derivative as follows:
ėA
Od 1 Od Od
A A A A O O A O O A A
E = Ė = ė2 = R E = R R E = OΩA EA (2.45)
dt dt dt
ėA
3
Here we have introduced the angular velocity matrix for the rotation of the basis EA
with respect to the fixed basis EO
Od
Ω ≡
O A A O
R O A
R . (2.46)
dt
2.5. INFINITESIMAL ROTATIONS AND ANGULAR VELOCITY 31
We now show that this is an anti-symmetric matrix. We denote the three by three unit
matrix (with ones on the diagonal) by 1 and the three by three null matrix, with all
elements zero, by 0. We can now write, using equation 2.16
Od Od Od Od
A OO A A O O A A O O A
0= 1= ( R R )= R R + R R (2.47)
dt dt dt dt
so that we have shown
Od Od
A O
R R = − ARO
O A O A
R . (2.48)
dt dt
But if we use the rule (AB)T = BT AT for the transpose of a matrix product, we find
that T T
Od Od Od
A O O A O A T A O A O O A
R R =( R ) R = R R . (2.49)
dt dt dt
We thus find that T
Od Od
A O
R R =−
O A A O
R O A
R (2.50)
dt dt
or, equivalently,
Ω = −( OΩA )T
O A
(2.51)
and this proves that the angular velocity matrix is anti-symmetric.
Just as in equation 2.43 above we can take the three elements of the anti-symmetric
matrix to define a corresponding vector. The vector corresponding to OΩA is the angular
velocity vector Oω A
0 Oω A
3 −Oω2A
Oω A
Ω = −Oω3A
O A
0 1 (2.52)
Oω A
2 −Oω A
1 0
and this matrix contains the components of this vector in the basis EA . Using this
vector we can now write the result of equation 2.45 in the form
Od ėA
1
Oω A × eA1
Oω A × eA
EA = ėA = ω ×E .
O A A O A A
2 = Ω E = 2 (2.53)
dt Oω A × eA
ėA
3 3
This equation should be compared to equation 2.44. It then shows that the small
change in the basis EA during the time dt is given by dEA = Oω A dt × EA so that the
basis is rotated by the infinitesimal rotation vector
δφ = O
ω A dt. (2.54)
One should note, however that the angular velocity vector is not the derivative of the
rotation vector for finite rotation angles φ.
Example 2.2 Calculate the angular velocity matrix and vector of the rotation matrix
cos ψ sin ψ 0
R = R3 (ψ) = − sin ψ
A O
cos ψ 0 . (2.55)
0 0 1
calculated in the example 2.1.
Solution: Using the definition 2.46 we get
O O cos ψ sin ψ 0 cos ψ − sin ψ 0
d d
O A
Ω = A O
R R =
O A
− sin ψ cos ψ 0 sin ψ cos ψ 0 (2.56)
dt dt
0 0 1 0 0 1
−ψ̇ sin ψ ψ̇ cos ψ 0 cos ψ − sin ψ 0 0 ψ̇ 0
= −ψ̇ cos ψ −ψ̇ sin ψ 0 sin ψ cos ψ 0 = −ψ̇ 0 0 . (2.57)
0 0 0 0 0 1 0 0 0
32 CHAPTER 2. KINEMATICS OF RIGID BODIES
which thus is parallel to the axis of rotation. In this example the axis of rotation is fixed,
but this is a general result; the angular velocity vector is parallel to the instantaneous axis of
rotation as indicated by equation 2.54. ✷
since in this basis only the components are time-dependent. If we now assume that the
reference frame A rotates with respect to the observer frame O, then when taking the
time derivative of the vector, as measured in O, one must take account of the fact that
also the basis EA is timedependent. We get
Od Od d A A Od Ad
r= (rA EA ) = ( r ) E + rA EA = r + rA OΩA EA . (2.60)
dt dt dt dt dt
connection Use of this and equation 2.53 leads to the important result
between the time
derivatives of a Od Ad
vector in r= r+ ω A × r.
O
(2.61)
dt dt
relatively
rotating reference Note that this is a vector equation and that it therefore is independent of the bases
frames
that are used. The components of the four vectors appearing in this equation may be
with respect to any suitable basis. When the basis is changed the components change
according to equation 2.20.
respect to the Euler angle rotated frame A given in equation 2.34 and figure 2.6. The
construction of this frame by means of a sequence of frames having simple rotations
relative to each other makes it easy to find its angular velocity. Generalizing the formula
2.63 to two intermediate frames, C and B, gives
O
ω A = Oω C + Cω B + Bω A . (2.64)
Since the relative rotation of these frames are of the simple type studied in example
2.2 reference to figure 2.6 immediately gives
O
ω A = ψ̇ eC3 + θ̇ eB1 + ϕ̇ eA3 . (2.65)
Using
eC
3 = sin θ sin ϕ eA A A
1 + sin θ cos ϕ e2 + cos θ e3 , (2.66)
eB
1 = cos ϕ eA
1 − sin ϕ eA
2, (2.67)
this gives
O
ω A = ψ̇ (sin θ sin ϕ eA1 + sin θ cos ϕ eA2 + cos θ eA3 ) + (2.68)
θ̇ (cos ϕ eA
1 − sin ϕ eA
2) + ϕ̇ eA
3
for the components of the angular velocity in the body fixed basis EA . We can write
this in the form
O A
ω1 = ψ̇ sin θ sin ϕ + θ̇ cos ϕ,
O A
ω2 = ψ̇ sin θ cos ϕ − θ̇ sin ϕ, (2.69)
O A
ω3 = ψ̇ cos θ + ϕ̇.
ri = OA + AP i = rA + ai = rO O A A
A E + ai E . (2.70)
In this expression the lengths of the vectors ai are constant since they are vectors
between points of a rigid body. The components, aA i , of the vectors ai with respect to
A
the basis E , must also be constant since both follow the body in its rotation. Thus
we can write the positions of the particles of the body in the form
The time-dependence of all N particles of the body is thus known once one knows
the time-dependence of the three components rO A (t) of the vector rA and the time-
dependence of the three parameters, φ(t), of the rotation matrix ARO from the observer
fixed basis EO to the body fixed basis EA . These parameters may be chosen as the three
components of the rotation vector or the three Euler angles or by other methods that
we have not treated here.
34 CHAPTER 2. KINEMATICS OF RIGID BODIES
vB = vA + O
ω A × AB. (2.73)
As long as one deals with a single rigid body moving with respect to a given fixed
reference frame, as will mostly be the case, there is no need for the superscripts O
and A on the angular velocity vector; we then simply write ω for the angular velocity
connection vector of the body. The above formula now takes the simple form
formula for
velocities in rigid vB = vA + BA × ω (2.74)
body
where we have used AB = −BA. This formula is analogous to the connection formulae
for moments (MB = MA + BA × F) and for angular momenta (LB = LA + BA × p).
Once the angular velocity vector and the velocity of one point are known this formula
gives the velocities of all other points of the body.
vA = vA
+ vA⊥ (2.75)
where
vA
= (vA · eω ) eω (2.76)
is the vector component of vA parallel to the angular velocity vector ω and where vA⊥
is the vector component perpendicular to the angular velocity vector (see figure 2.8).
instantaneous If we now choose the point C such that
axis of rotation
goes through the AC = (ω × vA )/ω 2 (2.77)
point C
we find algebraically, using 2.74 and the vector triple product, that
Figure 2.8: In this figure one sees the construction of the point C in the plane through A
perpendicular to the angular velocity vector ω . The velocity of the point C is parallel to ω
and so are the velocities of all points of the body, or rigidly connected to the body, on the axis
defined by C and ω . This axis is therefore the instantaneous axis of rotation.
Example 2.3 The position vectors and velocities of the three points P 1 , P 2 , and P 3 of a
rigid body have, at some instant of time, been found to be
v2 = v1 + ω × (r2 − r1 ), (2.85)
v3 = v1 + ω × (r3 − r1 ). (2.86)
Writing out the components of these equations explicitly we get the six equations
1v = 1v + [ωy · 0 − ωz · 0]#
0v = 1v + [ωz · 1 − ωx · 0]# (2.88)
−3v = −3v + [ωx · 0 − ωy · 1]#
The third of these gives ωx = 1 v , the sixth gives ωy = 0 v , and the fifth ωz = −1 v . Thus we
get
ω = (1, 0, −1) v# (2.89)
B
A
B
Figure 2.9: Graphic construction of the instantaneous centre of zero velocity C with the help
of two velocities of the body at points A and B. The diagram on the left is used when the
velocity vectors are parallel. The one on the right if they are not.
is not at a fixed point of space, it may move and have some velocity. It is the point of
the body, or rigidly connected to the body, which is at C at the given time which has
zero velocity. The point C of the body really has zero velocity but, in general, it will
have non-zero acceleration so it need not remain at rest.
When a wheel rotates around a fixed axis the instantaneous centre C is at rest at
the fixed point where the axis intersects the wheel. When a wheel rolls without slipping
on the ground the instantaneous centre is at the point of contact with the ground. The
material of the wheel must be at rest at that point since the wheel is not slipping and
the ground is at rest.
If we write the connection formula 2.74 for B= C, with vC = 0, as one of the points
we get (in the case of plane motion)
vA = AC × ω . (2.96)
This formula tells us that the velocity of a point is always perpendicular to the line from
the instantaneous centre C to the point, and that the magnitude of the velocity grows
linearly with the distance from C (vA = |AC|ω). These two facts lead to convenient
graphic methods for finding C from two known velocities of the body. This is illustrated
in figure 2.9.
Consider the plane motion of a rigid body and attach one coordinate system to the
body and let another coordinate system be fixed in space. The instantaneous centre
C will then trace out (plane) curves with time in these two systems. The curve in the
body fixed system is called the polehode curve and the curve in the space fixed system is
called the herpolehode curve. At a given instant of time these two curves touch (meet)
at the point where C is at that instant. The touching point of the polehode curve is
then at rest since it coincides with C while the herpolehode curve is fixed in space so
all its point are at rest. This means that the plane motion of the rigid body can be
viewed as the rolling without slipping of the polehode curve on the herpolehode curve.
As an example one might consider a circular wheel which rolls without slipping along a
straight line. Then the polehode curve is then the circle which defines the circumference
of the wheel and the herpolehode curve is the straight line.
38 CHAPTER 2. KINEMATICS OF RIGID BODIES
B
K
O
A C
Figure 2.10: The figure on the left refers to problem 2.4. The arrow on the cog-wheel with
centre at P must always point upwards.
Figure 2.11: The figure on the right refers to problem 2.5. Six identical cylindrical rollers are
connected by the rods AB, BC, and CA, so that they can rotate about their central axes and
roll on each other.
2.7 Problems
Problem 2.1 Use equation 2.65 (on page 33) to find the components of the angular
velocity vector Oω A along the fixed basis vectors EO .
Problem 2.2 By making measurements on two stereographic photographs of a rigid
body taken at times t and t + ∆t the position and velocity vectors of three points
P 1 , P 2 , P 3 , fixed on the body, have been found to have components
r1 = (1, 0, 0)#, v1 = (0, 1, 0)v,
r2 = (0, 2, 3)#, v2 = (0, −5, 4)v,
r3 = (2, 0, 1)#, v3 = (2, 3, −2)v,
relative to a coordinate system fixed in the body. The velocities are relative to a
reference frame in which the camera is fixed but their components have been determined
by projection to the body fixed basis. Find the components of the velocity of the origin
of the body fixed system and the components of the angular velocity vector of the body
in this body fixed system.
Problem 2.3 A homogeneous circular disc of mass m and radius r is rolling and sliding
in a vertical position along a straight line on a horizontal table. The uppermost point
of the disc has speed v1 and the geometric contact point with the table has speed v2
(note that this point is not a material point of the disc).
a) Find the angular velocity of the disc.
b) Calculate the angular momentum of the disc with respect to the geometric contact
point.
Problem 2.4 Three cog-wheels are connected as shown in figure 2.10. The wheel with
centre at the fixed point O has radius ro and is fixed so that it can’t rotate. A second
wheel of radius rk can roll on the fixed wheel and a third cog-wheel of radius rp can roll
on the second wheel. The centres of the three wheels are connected by an arm which
keep them on a straight line at fixed distances (such that they always touch). When
the arm is rotated an arrow painted on the outermost (third) wheel is required to point
upwards at all times. What relation between the radii ro , rk , and rp must hold if this
is to be the case?
2.7. PROBLEMS 39
Figure 2.12: This figure refers to problem 2.7. Two different positions of the ladder sliding
down are shown. The direction of the velocities of the end points of the ladder are indicated.
Problem 2.5 Six identical cylindrical rollers of radius r are connected by three rods
that keep them in contact with parallel axes so that they can roll on each other. The
six rollers form a pyramid with three at the base, see figure 2.11. Rods connect their
central axes and two, AB and BC, have length 4r while the one along the base, AC,
has length 6r. Use graphical methods to determine the instantaneous centres of zero
velocity for all six cylinders when the pyramid rolls to the right with speed v (with no
slipping anywhere).
Problem 2.6 The position vectors and velocities of the three points P 0 , P 1 , and P 2
of a rigid body have, at some instant of time, been found to be
Problem 2.7 A ladder of length # has been erected against a vertical wall. The
coefficient of friction against the horizontal floor is not large enough to keep the ladder
in equilibrium so it slides down as shown in figure 2.12. Determine the position of the
instantaneous centre of zero velocity C and find the equation for the curve described
by C when the ladder slides down,
a) with respect to a fixed coordinate system (the herpolehode curve), and
b) with respect to a coordinate system fixed in the ladder (the polehode curve).
c) Investigate how the two curves of a and b move relative to each other as the ladder
slides down.
40 CHAPTER 2. KINEMATICS OF RIGID BODIES
Since eB O O
1 = cos ψ e1 + sin ψ e2 , collecting the components gives the result
O A
ωx = ϕ̇ sin θ sin ψ + θ̇ cos ψ,
O A
ωy = −ϕ̇ sin θ cos ψ + θ̇ sin ψ, (2.97)
O A
ωz = ϕ̇ cos θ + ψ̇.
for the fixed frame components of the angular velocity in terms of Euler angles.
Answer 2.2 Use the connection formula 2.74 twice for the pairs P 1 P 2 and P 1 P 3 . This
gives six equations for the three unknowns ωx , ωy , ωz . Using these one finds
v
ω = (1, 2, 3)
#
for the angular velocity vector.
The connection formula also gives
vO = v1 + OP 1 × ω = v1 + r1 × ω
and use of this gives the velocity of the body fixed origin
vO = (0, −2, 2)v
when use is made of the components of ω determined above.
Answer 2.3 The crucial point to notice is that the speed of the geometric contact point
is the speed of the centre of the disc. This is best seen by going to a reference frame
that moves with the same velocity as the geometric contact point. In such a reference
frame the disc will only rotate and thus its centre will be at rest. Consequently the
centre must move with the speed of this reference frame i.e. the speed of the geometric
contact point. The answers are:
a) (v1 − v2 )/r.
b) 12 mr(v1 + v2 ).
Answer 2.4 Use the fact that the outermost wheel must have purely translational
velocity so that all its points have the same velocity. One finds that it is necessary that
ro = rp while rk is arbitrary.
Answer 2.5 See figure 2.13. The bottom rollers must have instantaneous centres of
zero velocity at the floor since they are assumed to roll without slipping. At the contact
points the rollers must have equal velocities for the same reason. The direction of the
velocity at the contact point with a roller of the second layer can be found by drawing a
line (dashed in the figure) from the instantaneous centre of zero velocity at the floor to
the contact point. The (common) velocity at the contact point must be perpendicular
to this line. The centres of the rollers must all have the common translational velocity
of the whole pyramid. By drawing lines perpendicular to these centre velocities (dashed
vertical lines in the figure) we get lines that have to go to the instantaneous centres
of zero velocity. In this way one finds that this point is at the top of the cylinders of
the second layer. By a line starting at this point and going through the contact point
with the top cylinder one gets the velocity of the contact point as a perpendicular to it.
Finally one then gets that the instantaneous centre of zero velocity of the top cylinder
is at its lowest point.
2.8. HINTS AND ANSWERS 41
Figure 2.13: This figure shows the instantaneous centres of zero velocity of three representative
rollers as black dots and refers to answer 2.5. The dashed lines are lines that are perpendicular
to the velocity vectors of certain points with known velocities.
Answer 2.6 Use the connection formula for velocities in a rigid body. The answers
should be:
a) ζ = −12,
b) v3 = − 54 (1, 7, 1)v.
Answer 2.7
a) The construction of the point C, the instantaneous centre of zero velocity, is indicated
graphically in figure 2.14. It is at the intersection of the two dashed lines, since these
are perpendicular to the velocities of the end points of the ladder.
If we denote the angle between the ladder and the floor α (this angle is π/2 when
the ladder is vertical and 0 when it is lying on the floor) we easily get that the position
of the point C is given by
OC(α) = #(cos α eO O
x + sin α ey ) (2.98)
where O is at the intersection of the floor and the wall and the basis vector eO x points
O
along the floor while ey points vertically upwards, see figure 2.14. # is the length of
the ladder. When α ∈ [0, π/2] this is the equation for a quarter circle with centre at O
and radius #.
b) To get an equation for the position of C with respect to a system fixed in the
ladder we take the origin at the middle of the ladder and denote it A. We let the basis
vector eA A
x point down along the ladder and ey be perpendicular to it. Note that the
point A is always halfway between O and C. Therefore
#
OA(α) = AC(α) = (cos α eO O
x + sin α ey ). (2.99)
2
We now wish to express the vector AC in terms of the A-basis. In order to do this we
use the expressions
eA x − sin α ey ,
= cos α eO O
x (2.100)
eA
y = sin α eO
x + cos α eO
y, (2.101)
42 CHAPTER 2. KINEMATICS OF RIGID BODIES
A
C
e y
eyO A exA
exO α
Figure 2.14: This figure shows the construction of the instantaneous centre of zero velocity C
for the ladder of answer 2.7. It also shows the curves that C traces out relative to a fixed system
and relative to a system fixed in the ladder when the ladder slides down. The former curve is
the herpolehode curve and is a quarter circle while the latter curve, which is a half circle, is the
polehode curve.
which can be read off from figure 2.14. An expression for the position vector of the
point C in the A-system can now be found using
# #
AC(α) = (AC · eA x ) ex + (AC · ey ) ey = (cos α − sin α) ex + (2 sin α cos α) ey .
A A A 2 2 A A
2 2
(2.102)
Standard formulae for the trigonometric functions can then be used to simplify this to
the final form
#
AC(α) = [cos(2α) eA A
x + sin(2α) ey ]. (2.103)
2
When the angle α varies from zero to π/2 this vector clearly describes a half circle with
radius #/2. This half circle is the polehode curve.
c) From figure 2.14 it should be clear that when the ladder moves and the angle
α varies the two curves move as if the polehode curve (the half circle attached to the
ladder) rolls, without slipping, on the herpolehode curve (the fixed quarter circle). This
is thus in agreement with the general theory of the plane motion of the rigid body.
Chapter 3
This chapter generalizes the definitions of the concepts of power, kinetic energy, work etc
to systems of particles. We discuss the concept of conservative force and its connection
with elasticity. Special attention is given to the special formulae obtained for rigid
bodies. We discuss which forces that do work and which don’t and the role of the law
of conservation of energy
N
N
P = Pk = Fk · ṙk . (3.1)
k=1 k=1
The kinetic energy T for a particle system is, likewise, defined as the sum of the kinetic
energies, Tk = 12 mk vk2 , of the individual particles kinetic energy
N
N
1
T = Tk = mk ṙk · ṙk . (3.2)
k=1 k=1
2
Since
1
Ṫk = mk (r̈k · ṙk + ṙk · r̈k ) = mk r̈k · ṙ = Fk · ṙk , (3.3)
2
where we have used mk r̈k = Fk , we find immediately, that, with these definitions, the
power and the kinetic energy obey the relationship
P = Ṫ (3.4)
in the same way as for a single particle.
We now split the velocities vk of the system into centre of mass velocity vG and
velocity vk relative to the centre of mass, as we did in subsection 1.1.1, so that vk =
vG + vk , and use this in the kinetic energy expression. We get
N
1
N
1
T = mk vk · vk = mk (vG + vk ) · (vG + vk ) (3.5)
k=1
2 k=1
2
1
N
N N
1
= vG · vG ( mk ) + vG · ( mk vk ) + mk vk · vk . (3.6)
2 k=1 k=1 k=1
2
43
44 CHAPTER 3. ENERGY OF PARTICLE SYSTEMS AND BODIES
Because of the centre of mass constraint the sum in the middle term of the last line
the two parts of here, is zero (see equation 1.10), and we thus get
the kinetic energy
or, König’s 1 N
1 1
theorem T = mvG · vG + mk vk · vk = mvG2 + T . (3.7)
2 k=1
2 2
The kinetic energy can thus be thought of as having two parts, one due to the translation
of the centre of mass and the rest which comes from the internal motions of the system
relative to its own centre of mass system. In Germanic literature this formula is often
referred to as ‘König’s’ theorem.
When this is inserted into the expression T for the internal part of the kinetic energy
we get
N
1 N
1
T = mk vk · vk = mk (ω × rk ) · (ω × rk ). (3.10)
k=1
2 k=1
2
To evaluate this further we must find a formula for the scalar product and this can be
done as follows
1 N
1 N
T = ( mk rk · rk )(ω · ω ) − mk (ω · rk )2 . (3.14)
2 k=1 2 k=1
and the other moments and products of inertia are obtained by cyclic replacements
x → y → z → x. We see, first of all, that if the body does not rotate then T = 0
since ω = 0. This is the reason why a rigid body with only translational motion can
be treated like a particle. One should also note that the moments and products of
3.1. POWER AND KINETIC ENERGY 45
inertia (with respect to axes through the centre of mass) will, in general, depend on
time unless the basis vectors (or equivalently, the direction of the coordinates axes) are
taken as fixed in the body. When this is done these quantities become constant.
In case one point C of the rigid body has zero velocity, permanently due to some
constraint, or instantaneously as the instantaneous centre of zero velocity that exists
for plane motion, we can write the velocities of the particles as
vk = ω × CP k (3.18)
according to the connection formula. If we put the origin of the coordinate system at
C this takes the form
vk = ω × rk . (3.19)
The (complete) kinetic energy T now takes the form
1 N N
1
T = mk vk · vk = mk (ω × rk ) · (ω × rk ). (3.20)
2 k=1 k=1
2
This is formula is identical in form to 3.10 so the same kind of calculation gives kinetic energy of
rigid body with
1 1 1
T = JxC ωx2 + JyC ωy2 + JzC ωz2 − DxyC C
ωx ωy − Dxz C
ωx ωz − Dyz ωy ωz . (3.21) pure rotation
2 2 2
This is thus the kinetic energy of a rigid body whose motion is a pure rotation around
the point C. The moments and products of inertia must now be calculated with respect
to axes through the point C. As above, they are, in general, constants only if the
components of the vectors ω and rk are taken with respect to a basis that rotates with
the body.
where ez has been chosen in this fixed direction. The two formulae for the kinetic
energy, 3.15 and 3.21 above, reduce to simple forms. The internal energy becomes
1
T = JzG ϕ̇2 . (3.26)
2
Pure rotation around C gives the kinetic energy
1
T = JzC ϕ̇2 . (3.27)
2
It can be instructive to derive this last result directly from the definition using
cylindrical coordinates. Since, with the origin at C,
we get
N
1
N
1
T = mk vk · vk = mk (ϕ̇ρk eϕ ) · (ϕ̇ρk eϕ ) (3.29)
k=1
2 k=1
2
N
1 1 N
1
= mk ϕ̇2 ρ2k = ( mk ρ2k )ϕ̇2 = JzC ϕ̇2 . (3.30)
k=1
2 2 k=1 2
There is a connection between formulae 3.26 and 3.27 which is derived in the example
below.
Example 3.1 Use the parallel axis theorem for moments of inertia to show that in the case
of plane motion
1 C 2 1 1
J ϕ̇ = mvG2 + JzG ϕ̇2 . (3.31)
2 z 2 2
and thus verify equation 3.7 (T = 12 mvG2 + T ). The point C is assumed to be in the same
xy-plane as G.
Solution: The parallel axis theorem gives us
so we get
1 C 2 1 1 1
J ϕ̇ = (JzG + m|CG|2 )ϕ̇2 = JzG ϕ̇2 + m(|CG|ϕ̇)2 . (3.33)
2 z 2 2 2
But the connection formula for velocities gives us
vG = ω × CG (3.34)
and in our plane situation this means that |vG | = |ω ||CG| = |ϕ̇||CG| so that
1 C 2 1 1
Jz ϕ̇ = JzG ϕ̇2 + mvG2 (3.35)
2 2 2
and this is what was to be shown. Note how the parallel axis theorem and the connection
formula for the velocities act together to give this result. ✷
3.2. THE WORK OF EXTERNAL AND INTERNAL FORCES 47
so the work done on the system is equal to the increase in kinetic energy.
Using the definition, equation 3.1, we can rewrite 3.36 in the form
t2
N N
W1,2 = ( Fk · ṙk ) dt = Fk · drk , (3.38)
t1 k=1 k=1 Ck (1,2)
where we have introduced the line integral Ck (1,2) Fk · drk along the path Ck (1, 2) : t ∈
[t1 , t2 ] → rk (t) followed by particle k between the points rk (t1 ) and rk (t2 ) during the
time from t1 to t2 . The work done on the particle system can thus also be expressed as
such a sum of the work on the individual particles expressed in terms of line integrals.
To simplify the notation we shall now consider this sum to be a single line integral along
the curve (x1 (t), y1 (t), z1 (t), x2 (t), . . . , xN (t), yN (t), zN (t)) through the 3N -dimensional
configuration space of the system. We denote this 3N -dimensional curve, traversed
between t1 and t2 , by C N (1, 2) and thus write the work on the particle system as the
single configuration space line integral work as line
integral along
N configuration
W1,2 = Fk · drk . (3.39) space curve
C N (1,2) k=1
where the superscript ‘e’ stands for external and ‘i’ stands for internal, and the
We assume that Newton’s third law fjk = −fkj holds. The total work of formula 3.39
can now be analyzed into external and internal contributions as follows
N
N
N
W1,2 = Fek · drk + fkj · drk ≡ W1,2
e i
+ W1,2 . (3.43)
C N (1,2) k=1 C N (1,2) k=1 j=1
Here we have used only the ‘anti-symmetry’ part of Newton’s third law. We now express
the working external and internal forces as sums of a conservative part, which is the
negative gradient of a potential energy, and a non-conservative part:
We can now split the external work into contributions from decrease in potential energy
and non-conservative work as follows:
N
N
e
W1,2 = −[∇k Φk (rk )] · drk + k · drk
Fen (3.49)
C N (1,2) k=1 C N (1,2) k=1
N
= −[Φk (rk (t2 )) − Φk (rk (t1 ))] + W1,2
en
(3.50)
k=1
N
Φe (i) ≡ Φk (rk (ti )), (3.52)
k=1
1
N
Φi (i) ≡ Φkj (rkj (ti )). (3.53)
2 k,j=1
If we now collect all results we have the following expression for the work done on a
the four types of particle system
work
e
W1,2 = W1,2 i
+ W1,2 = {−[Φe (2) − Φe (1)] + W1,2
en
} + {−[Φi (2) − Φi (1)] + W1,2
in
}. (3.54)
3.2. THE WORK OF EXTERNAL AND INTERNAL FORCES 49
The total work can thus be analyzed into these four contributions from external and
internal, conservative and non-conservative forces. The parts of the work which depend
only on the change in total potential energies have the fundamental property that they
only depend on the positions of the particles, i.e. on the position of the system in
configuration space, and not on the path followed between the positions. If the system
returns to the original position no net work has been done by the conservative forces.
The internal total potential energy Φi has the interesting property that it only depends
on the relative positions and thus not on the position and orientation of the system
as a whole. For rigid motions, translations and rotations, Φi does not change; only
changes in the shape of the system changes Φi . This part of the energy of a body is
the elastic energy. It is large when the body has been deformed but can be recovered
as mechanical work when the body returns to its un-loaded shape.
Example 3.2 Calculate the external total potential energy Φe due to the weights of the
particles of the system.
Solution: If we denote the acceleration due to gravity by g we have that
N N
mk rk
Φe = −mk g · rk = −mg · k=1
N
= −mg · rG = mgh. (3.56)
k=1 k=1 mk
The total potential energy is thus completely determined by the total mass m, the acceleration
due to gravity g, and the vertical projection of the centre of mass position vector. ✷
Example 3.3 Find the internal total potential energy Φi of an elastic rubber-band assuming
that the force required to stretch the band the length ∆# is given by F = k ∆#, where k is a
constant.
Solution: The work done by the internal forces during the stretching is easily found by
integration and the result is that
1 2
i
Φ = 2 k(∆#) for ∆# > 0
. (3.57)
0 for ∆# ≤ 0
Only stretching leads to internal work since the flexible rubber-band does not take up compres-
sive forces. ✷
We have found two different expressions for the total work done on a particle system,
namely in terms of kinetic energy in equation 3.37, and that of equation 3.54. If these
two expressions are set equal we can rearrange the equation and find that
One defines the total mechanical energy Em to be the kinetic plus the total potential
energies: mechanical
Em ≡ T + Φe + Φi . (3.59) energy
The previous formula now states that the change in the mechanical energy is equal
to the work of, external and internal, non-conservative forces. One can often arrange
the external non-conservative forces to be small or negligible. Deformations of bod-
ies, however, always entail some amount of non-conservative internal work. The best
chances for mechanical energy conservation (Em =const.) therefore come about when
deformations are negligible. In the following subsection we will find that the internal
work is zero for rigid bodies, but one should remember that complete rigidity is an
idealization which is only approximated by nature.
50 CHAPTER 3. ENERGY OF PARTICLE SYSTEMS AND BODIES
Example 3.4 Consider a pole-vaulter with an elastic glass-fiber pole as a particle system and
analyze the work on the system in terms of the four kinds of work in equation 3.54.
Solution: When the athlete starts his run work is done by his muscles and this must be
considered as coming from internal non-conservative forces so we have the type W in . At the
end of the run this work has produced a lot of kinetic energy. The pole-vaulter now transforms
this kinetic energy plus some more muscle work from the arms into elastic deformation energy
of the pole i.e. type Φi . The pole then straightens out again and raises the athlete vertically
upwards. The internal potential energy of the pole then is transformed into external potential
energy Φe of the gravitational field. At the highest point of the jump all energy has become
external potential energy. When he falls down it again becomes kinetic energy. This kinetic
energy is then destroyed by negative work by (mainly) external non-conservative forces W en
from the shock absorber that bolsters his impact with the ground. ✷
where δ φ is the infinitesimal rotation vector. We now insert this in the expression for
the external work as given by equation 3.43, and get
N
N
e
W1,2 = Fek · drk = Fek · (drA + (rA −rk )×δ φ) = (3.61)
C N (1,2) k=1 C N (1,2) k=1
N
{Fek · drA + Fek · [(rA −rk )×δ φ]} = (Fe · drA + MeA · δ φ). (3.62)
C N (1,2) k=1 C N (1,2)
Here we have rearranged the scalar triple product and used the definitions Fe =
N e
k=1 Fk and
N
MeA = (rk − rA ) × Fek . (3.63)
k=1
total internal force and moment are zero (see equations 1.23 and 1.32), we get that the
work of the internal forces
i
W1,2 =0 for a rigid body. (3.64)
work done on Summarizing we have now found that the work on a rigid body is given by
rigid body by
external force
and moment W1,2 = (Fe · drA + MeA · δ φ), (3.65)
C N (1,2)
i.e. as a line-integral depending on the external force and the external moment.
It can be instructive to calculate the internal work directly in terms of inter-particle
forces to see in more detail why (or when) it is zero. To do this we need an expression
3.2. THE WORK OF EXTERNAL AND INTERNAL FORCES 51
for drkj = d(rk − rj ), and to get that we put rA = rj in the connection formula 3.60.
Some rearrangement then gives
We now use this in formula 3.46 for the internal work and get
1 N
1 N
i
W1,2 = fkj · drkj = fkj · (δ φ × rkj ). (3.67)
C N (1,2) 2 k,j=1 C N (1,2) 2 k,j=1
Just as in subsection 1.1.2 we now see that this is manifestly zero only if assume that the
internal forces are parallel to the inter-particle vectors: fkj rkj since then fkj ⊥ drkj .
Now that we have established that the internal forces do not contribute to rigid body
work we will not bother with the superscripts ‘e’ on forces and moments any more.
The notation C N (1, 2) which we introduced for the path through configuration space
of the N -particle system is not well suited for the rigid body case, since in this case the
number of particles is immaterial; only the change in the six degrees of freedom rA , φ
with time matters. The first part of the work in equation 3.65 is thus an ordinary line
integral along the curve on which the point A moves through space, but the second
part is a curve in the space of orientation (rotation) parameters (e.g. Euler angles). If
we take the centre of mass as base point we can thus express the work in the form
W1,2 = F · drG + MG · δ φ. (3.68)
C G (1,2) C R (1,2)
If the external force is the gradient of a total potential energy, as is the case for gravity,
the first of these two parts can be expressed with the help of decrease in potential
energy in the way discussed above. Also, if there is only gravity, then by the definition
of centre of mass as the point of application of the resultant, we have MG = 0, so that
the second, rotational part of the work gives zero. One then gets
grav
W1,2 = −∇G Φ(rG ) · drG = −[Φ(rG (t2 )) − Φ(rG (t1 ))] (3.69)
C G (1,2)
for the work of gravity on a rigid body, the same as for any particle system. Here
Φ = Φe is given by equation 3.56.
Should the moment, on the other hand, be non-zero there is, in general, no hope of
representing the rotational part of the work with a ‘rotational’ potential energy since
the integration element δ φ is not the differential of a vector. The exception is when
there is a fixed direction, ez , of the rotation axis. In this case one has δ φ = dφ ez so
the rotational part of the work is simply
rot
W1,2 = MAz dφ. (3.70)
C R (1,2)
Should one be able to find a model for the moment in which it is a function of the
rotation angle φ alone: MAz = MAz (φ) then there is also a rotational potential energy
given by
φ
Φrot (φ) = − MAz (φ ) dφ . (3.71)
Example 3.5 A homogeneous circular disc of mass m and radius R is attached horizontally
to a thin vertical staff at its mid-point. See figure 3.1. The vertical staff has the property
that when it is twisted an angle φ it responds with a moment in the opposite direction which
is proportional to the amount of twisting: M (φ) = −τ0 φ. Determine the period for twisting
52 CHAPTER 3. ENERGY OF PARTICLE SYSTEMS AND BODIES
r
φ
h
x
φ
X
α
Figure 3.1: The torsion pendulum of example 3.5 is shown in the figure on the left. In the
figure the disc has been twisted an angle φ. The line fixed on the disc is assumed to be parallel
to be fixed direction of the arrow at equilibrium, when the vertical staff is not twisted.
Figure 3.2: The figure on the right shows a cylinder rolling down an incline which makes an
angle α with the horizontal. In example 3.6 the speed of the cylinder, after it has sunk a vertical
distance h, is calculated.
2π τ0 2τ0
Tp = = 2π = 2π . (3.76)
ω0 Jz mR2
In this example it was not particularly advantageous to use the conservation of mechanical
energy, but in other more complicated cases it can be very useful as we’ll see below. ✷
When rigid bodies are in contact they act with contact forces upon each other. Such
contact forces can be classified into three types: a) Normal forces, b) Static friction,
and c) Sliding (or kinetic) friction. Of these three only the last type, kinetic friction,
does any net work on the combined system of bodies in contact. We now discuss why
this is so.
First consider the normal force on a moving body from the surface of a fixed body.
By definition of normal force it acts perpendicularly to the possible direction of motion normal force does
of the moving body at the point of contact. The work dW = N · dr must then be no work
zero since the force vector N and the displacement vector dr are perpendicular. If
on the other hand both bodies move then the component of the motion along N will
give non-zero net work on the body on which N acts. Because of Newton’s third law,
however, a reaction force of equal magnitude but opposite direction must act on the
other body. Since this body has the same velocity component perpendicular to the
surface of contact as the other body, the net work of force and reaction force is zero.
The case of static friction forces is similar. When such a force acts from a body at static friction
rest on a moving body, the point of the moving body in contact must, by definition of does no work
static friction, be at rest. This means that the motion of the moving body must be a
rotation about the point of contact. The work done by the force must thus be zero since
the displacement vector dr at the point of contact is zero dr = 0. If both bodies are
moving the same reasoning as above, for normal forces, shows that the static friction
force and its reaction force on the other body, together produce zero net work.
The two non-working forces discussed above occur, for example, when there is rolling
without slipping. They also occur when there are smooth constraints, i.e. well lubricated
hinges, bearings, ball and socket joints, tracks to slide along etc. Should these devices
not be well lubricated kinetic friction will occur as non-conservative internal force (with
respect to the system as a whole) which does negative work. The mechanical energy
then decreases (dissipates) and becomes heat, vibration and noise.
Example 3.6 A homogeneous circular cylinder of mass m and radius r rolls without slipping
on an incline which makes an angle α with the horizontal. See figure 3.2 for the geometry.
Assume that the cylinder starts from rest. Find the centre of mass speed of the cylinder after
it has sunk the vertical distance h.
Solution: Since there is rolling without slipping the mechanical energy is conserved and only
the gravitational force does work on the rolling cylinder. In figure 3.2 the angle φ is the angle
that a line fixed on the cylinder has rotated from the initial position. Rolling without slipping
means that the length x traversed along the incline must be equal to the length rφ along the
circumference of the cylinder:
x = rφ. (3.77)
This is usually called the ‘rolling constraint’. We now write down the energy of the cylinder
using equations 3.7 and 3.26 and the expression 3.56 for the potential energy
1 1
E= mẋ2 + JG φ̇2 − mgx sin α. (3.78)
2 2
Here JG = 12 mr2 and vG = ẋ. If we use the rolling constraint we get φ̇ = ẋ/r. At the initial
position we have ẋ = 0 and x = 0 so the energy E = 0. Since h = x sin α we now get
1 11 2
0= mẋ2 + mr (ẋ/r)2 − mgh. (3.79)
2 22
From this one easily calculates
4gh
ẋ = (3.80)
3
and this is thus the desired speed. ✷
Example 3.7 A smooth stiff wire of mass 3m has been given the shape of a semi-circle with
radius r. The semi-circle hangs vertically down from a smooth horizontal axis through small
54 CHAPTER 3. ENERGY OF PARTICLE SYSTEMS AND BODIES
loops at its end points. A small pearl of mass m can slide with negligible friction along the
semi-circular wire. The pearl is released from rest at one of the end points of the wire in contact
with the horizontal axis.
a) Show that the trajectory of the pearl is an ellipse.
b) Find the speed of the pearl in its lowest position.
Solution: a) Let the horizontal axis be the x-axis and place the origin at the initial position
of the mid-point between the ends of the semi-circular wire:
The position of the pearl is then: (x(0), y(0) ) = (r, 0). We choose the y-axis vertically down-
wards. The absence of external horizontal forces means that the x-coordinate of the center of
mass will be at rest and we get
1
xw (t) = [r − x(t)] (3.82)
3
between the x-coordinates. The fact that the pearl at (x, y) is on part of the circle with radius
r and mid-point (xw , 0) gives the following relationship
(x − xw )2 + y 2 = r2 . (3.83)
We now eliminate xw from this using the previous equation and get
1
[x − (r − x)]2 + y 2 = r2 . (3.84)
3
Some algebra leads to the expression
2 y 2
x − r/4
+ =1 (3.85)
3r/4 r
which, clearly, is the equation for an ellipse with center at (r/4, 0) and the pearl must move on
this ellipse.
b) Since there is no friction in this problem the only force that does work is gravity and
the energy is thus conserved. The wire cannot move vertically and must thus have constant
potential energy. We get the following expression for the total energy
1 1
m(ẋ2 + ẏ 2 ) + 3mẋ2w − mgy = E. (3.86)
2 2
Since initially everything is at rest and y = 0 we have E = 0. Differentiating the relationship
xw (t) = 13 [r − x(t)] above with respect to time gives ẋw = −ẋ/3. Inserting this into the energy
expression gives
2
1 1 ẋ
2 2
m(ẋ + ẏ ) + 3m − mgy = 0. (3.87)
2 2 9
At its lowest position the pearl has vertical velocity so ẏ = 0 and y = r. This now gives for the
speed v = ẋ at this position
2
1 v
m v2 + − mgr = 0 (3.88)
2 3
so we find that
3
v= gr. (3.89)
2
is the speed of the pearl at the lowest position. ✷
3.3. PROBLEMS 55
Figure 3.3: The figure refers to problem 3.3 and shows the steam roller and how the braking
force is applied to its front wheel.
3.3 Problems
Problem 3.1 Find the kinetic energy of the rolling and sliding circular disc of problem
2.3 (on page 38).
Problem 3.2 Consider the pyramid of rollers of problem 2.5 on page 39. Assume
that each cylindrical roller is homogeneous and of mass m, radius r and has speed v.
Calculate the kinetic energy of the pyramid
a) using König’s theorem, formula 3.7, and
b) using the existence of the instantaneous centre of zero velocity in the plane motion
of the rigid body.
Problem 3.3 In order to reduce the transport weight and braking distance of the
steam roller shown in figure 3.3 it is suggested that the wheels are made as hollow
cylindrical shells. These are then filled with water when the steam roller is to be used.
Assume that such a steam roller with empty wheels, has a braking distance of 2 m
at a speed of 5 km/h. The brake consists of a chock which is pressed against the front
wheel with a force of F = 125 · g N, the coefficient of (sliding) friction being 0.8 between
wheel and chock. The wheels have a diameter of 1 m and a width of 2 m. Calculate the
braking distance when the wheels are filled with
a) liquid water, and
b) ice.
Problem 3.4 One can specify the performance of a car engine by giving either the
maximum power Pmax that it can deliver, or its maximum moment (or torque) Mmax .
Since Mmax determines the acceleration of the car it is common to specify at what
number of revolutions (per unit time) nmax that this maximum torque is achieved.
Assume that the torque, M , is a known function, M (n), of the number of revolutions
n.
a) Derive a formula for the power as a function of n.
b) Show that the maximum power and the maximum torque cannot correspond to the
same n.
Problem 3.5 One part of a friction clutch rotates freely around its axis with angular
velocity ω as shown in figure 3.4. The moment of inertia of this part is J. It is suddenly
pressed with force P against an identical but initially non-rotating part. The coefficient
of friction between the two parts is f and the contact area is the plane annulus with
56 CHAPTER 3. ENERGY OF PARTICLE SYSTEMS AND BODIES
R ω ω
r
A C B
ω
Figure 3.4: A vertical cross section through the symmetry (rotation) axis of the friction clutch
of problem 3.5 is shown in the figure on the left.
Figure 3.5: The figure on the right refers to problem 3.6. Two identical discs rotate in opposite
directions around parallel axes through the ends of a light bar AB. The bar can rotate around
an axis through its midpoint C around a fixed axis parallel to those of the discs.
inner radius r and outer radius R. All other friction is negligible. Calculate
a) the final common angular velocity of the clutch, and
b) the total heat loss in the clutch.
Problem 3.6 Two identical homogeneous circular discs, of mass m and radius R,
rotate in opposite directions around parallel axes through the ends of a light bar AB,
see figure 3.5. The bar can rotate around a fixed axis, parallel to those of the discs,
through its midpoint C. The rotation axes are all perpendicular to the plane of the
figure 3.5, and the friction can be neglected in the bearings at B and C. Initially the
bar, which has length #, is at rest and the discs have angular velocity ω. Calculate the
heat loss in the bearing at A once the rotation with respect to this bearing has stopped.
3.4. HINTS AND ANSWERS 57
Answer
! 3.2 "
a) 6 2 mv 2 + 12 JG (v/r)2 = 92 mv 2 .
1
! "
1
b) 6 2 JC (v/r)
2 = 92 mv 2 , since the parallel axis theorem gives JC = JG + mr2 .
Equating the work done by the braking force F to the initial kinetic energy of the steam
roller we get
2
1 1 v
(m0 + 2me )v 2 + 2 Je = µF #e .
2 2 r
When the wheels contain water the steam roller can come to halt without dissipating
the rotational energy of the water. The extra kinetic energy that the brake force must
make zero is thus 2 12 mw v 2 . Using the previous result, this gives
µF #e + mw v 2 = µF #w
where #w is the braking distance with liquid water in the wheels. This gives
mw v 2
#w = #e +
µF
Putting numbers in we find that mw = π · 500 kg, v = (5/3.6) m/s, and µF = 0.8 · 125 ·
9.81 N so that the extra braking distance with water in is roughly 3 m. Since #e = 2 m
we thus find
a) answer #w = 5 m.
When the water is frozen to ice one must also dissipate the rotational energy
2
of the
water to brake the steam roller. The extra kinetic energy is then 2 12 Jw vr . In the
same way as above one then finds that
1 mw v 2
#i = #w +
2 µF
is the new braking distance. This thus adds an extra 1.5 m to give
b) answer #i = 6.5 m.
Answer 3.4
a) Use of
t
d
P = M dϕ = M ϕ̇
dt
58 CHAPTER 3. ENERGY OF PARTICLE SYSTEMS AND BODIES
Answer 3.5
Note that the details of the process between the initial and final state of motion is
irrelevant. the answers will thus not depend on P and f .
a) The final angular velocity is ω/2.
b) The heat loss is Jω 2 /4.
Answer 3.6
Use of Lz =const. gives the angular velocity of the bar after the braking. It is found
to be
R2
Ωfinal = − 2 ω,
# + R2
where the sign assumes that the disc at B has positive angular velocity before the
braking (as well as after). The heat loss is given by
1 R2
Tinitial − Tfinal = mR2 ω 2 1 − 2
4 # + R2
and is thus always positive. Note that when # → 0 the heat loss goes to zero. The
reason for this is that when # = 0 the bar will have no inertia and the disc at A can
tranfer its angular velocity to the bar with no loss of energy. Energy is then conserved.
Chapter 4
In this chapter we present the general form that the laws of linear and angular mo-
mentum take for rigid bodies. The law of angular momentum leads to Euler’s dynamic
equations for the rigid body. The concept of the inertia tensor is discussed and it is
stressed that the equations of motion should be given in such a way that the elements
of the inertia tensor are constant. Pendulums and reaction forces on rotating bodies
are also treated.
N
L= rk × mk vk . (4.1)
k=1
Here rk are the position vectors of the particles of the body with respect to the relevant
base point (G or C) and vk are the velocities of the particles. In the case of a fixed point,
C, these are absolute velocities and are given by equation 3.19. In the case of the centre
of mass we use velocities relative to the centre of mass system (so that we calculate
LG ) and these are given by equation 3.9. Both these equations for the velocities have
the form
vk = ω × rk . (4.2)
The angular momentum that we want to calculate will thus in both cases have the
algebraic form
N
L= rk × mk (ω × rk ) (4.3)
k=1
and this can be rewritten, using the formula for the triple vector product,
N
L= [ω mk (rk · rk ) − mk (ω · rk )rk ]. (4.4)
k=1
The three components of this equation now give the result: angular
momentum of
Lx = Jx ωx − Dxy ωy − Dxz ωz , rigid body
59
60 CHAPTER 4. DYNAMICS OF RIGID BODIES
One should remember that the elements of these matrices depend on the basis triad
used. There is, however, an invariant geometric relationship between the two vectors
L and ω expressed by this equation. We now investigate this.
and correspondingly for ω and its components w. The equation 4.6 now gives
L = LO EO = wO JO EO (4.8)
in the fixed (observer) basis triad O. The same formula can also be expressed in a body
fixed basis triad EA so that we also have
L = wA JA EA . (4.9)
We now make use of the formulas 2.10 and 2.20 to transform from the O to the A basis
with the rotation matrix ARO and its inverse (=transpose) ORA . The above equation
then gives
L = wO ORA JA ARO EO . (4.10)
If we now compare this with equation 4.8 we find that the matrix of moments and
transformation of products of inertia for the two sets of basis vectors must be related according to
the inertia tensor
under rotation JO = ORA JA ARO . (4.11)
Once the moments and products of inertia have been calculated for one set of axis
directions one can thus find the corresponding quantities for any other set of axes
simply by multiplying with the relevant rotation matrix according to this formula.
In the chapter on rigid body kinematics we found that to each rotation matrix
there corresponds a rotation operator R̂ which operates on vectors to give new rotated
vectors. The operator is a linear operator , i.e. it obeys R̂(λa + µb) = λR̂a + µR̂b.
Such linear operators are geometric ‘objects’, independent of any particular basis in
which their components are given, and they are sometimes called (second rank) tensors
or ‘dyads’.
Since equation 4.6 clearly defines a linear transformation from the components of
ω to L there is a corresponding linear operator Jˆ which transforms the vector ω to
the vector L
L = Jˆω = wA JA EA . (4.12)
This linear operator is thus what properly should be called the inertia tensor of the
body even though this name is often used for the matrix of its components in some
basis. Unlike the rotation operators more general linear operators change both the
4.1. THE ANGULAR MOMENTUM OF A RIGID BODY 61
direction and the length of the vector on which it operates. Nor is the rotating part of
the operation a rigid rotation; different vectors are rotated in different ways.
When a rigid body moves its inertia tensor will, in general, be constant only if
it is calculated with respect to a point fixed in the body, or rigidly connected to the
body, and with respect to axis directions fixed in the body. The requirement that the
axis directions are fixed in the body, however, does not specify any particular direc-
tions for these axes; they can still be chosen in an infinity of different ways differing
from each other by constant rotations. General mathematical theory of linear trans-
formations, however, indicates that when such transformations are represented by real
symmetric matrices there might exist unique mutually orthogonal eigen-vectors of the
transformation. These eigen-vectors eA i are defined by the equation
ˆ A = J eA
Je i = 1, 2, 3 (4.13)
i i i
and the numbers Ji are the eigen-values 1 . These eigen-vectors of the inertia tensor
define the principal axis directions of the body and the component matrix of the inertia
tensor in this principal basis is diagonal with the eigen-values along the diagonal:
Jx 0 0
J’ = 0 Jy 0 . (4.14)
0 0 Jz
These diagonal matrix elements are the principal moments of inertia of the body. In
what follows we will normally omit the primes on the principal moments of inertia and
simply write Jx etc.
The principal axes of a body can sometimes be found by means of the following two
rules:
(I.) Any plane of symmetry of a body is perpendicular to one of the principal axes.
(II.) A symmetry axis of a body is a principal axis. Any pair of axis perpendicular
to the symmetry axis will be principal axes and correspond to equal principal moments
of inertia.
A body for which two of the principal moments of inertia are equal is said to be a
symmetric top. The two eigen-vectors in the plane corresponding to these two moments
are then not unique; any pair will do. If all three moments of inertia are equal, the
inertia tensor does not single out any direction of the body as special; all vectors are
eigen-vectors. Such a body might be called a ‘spherical top’.
Example 4.1 A rigid body has mass m and principal moments of inertia Jx , Jy , and Jz .
Show, by explicit construction that there exists a rigid four particle system with the same mass
and inertia tensor.
Solution: Consider the four particle system in figure 4.1, where all particles have mass m/4
so that the total mass is m. The position vectors are given by
r1 = ( a, 0, c), (4.15)
r2 = (−a, 0, c), (4.16)
r3 = ( 0, b, −c), (4.17)
r4 = ( 0, −b, −c). (4.18)
The elements of the inertia tensor matrix for this system are
m
Dxy = mi xi yi = (a · 0 − a · 0 + 0 · b − 0 · b) = 0 (4.19)
4
m
Dxz = mi xi zi = (a · c − a · c − 0 · c − 0 · c) = 0 (4.20)
4
1
One sometimes finds that the terms ‘characteristic vectors’ and ‘characteristic values’ are used
instead.
62 CHAPTER 4. DYNAMICS OF RIGID BODIES
Y
a a
m/4 m/4
c
X
c m/4
b
m/4
Figure 4.1: This figure shows the four particle system for which the inertia tensor is calculated
in example 4.1. The total mass is m and the four identical particles are connected by light rigid
rods.
m
Dyz = (0 · c + 0 · c − b · c + b · c) = 0
mi y i zi = (4.21)
4
m
Jx = mi (yi2 + zi2 ) = (2b2 + 4c2 ) (4.22)
4
m
Jy = mi (xi + zi ) = (2a2 + 4c2 )
2 2
(4.23)
4
m
Jz = mi (x2i + yi2 ) = (2a2 + 2b2 ). (4.24)
4
Since the inertia tensor is diagonal the axes are the principal axes for this case. The principal
moments of inertia for this four particle system are thus given by
m 2 m 2 m 2
Jx = (b + 2c2 ), Jy = (a + 2c2 ), Jz = (a + b2 ). (4.25)
2 2 2
Since we consider the moments of inertia as given and the particle system geometry as unknown,
we must solve these equations for the distances a, b, and c in terms of Jx , Jy , and Jz . Some
algebra shows that
Jx + Jy − Jz = 2mc2 , (4.26)
Jz + Jx − Jy = mb2 , (4.27)
Jz − Jx + Jy = ma2 , (4.28)
so the our four particle system will have the desired inertia tensor provided the distances are
chosen as follows:
1 1 1
a= (Jz − Jx + Jy ), b= (Jz + Jx − Jy ), c= (Jx + Jy − Jz ). (4.29)
m m 2m
The four particles can thus be arranged to have any desired inertia tensor.
Should two principal moments of inertia be equal, Jx = Jy say, one finds that this means
that the two distances a and b must be equal: a = b. A body with two equal principal moments
of inertia is said
√ to be a symmetric top. All three principal moments of inertia will be equal if
a = b and a = 2c; in this case the inertia tensor is said to have spherical symmetry. The four
particles are then at the corners of a regular tetrahedron.
The fact that a three particle system cannot produce a general inertia tensor is best seen
from the fact that such a system always will be plane (the plane of the three particles). If this
plane is chosen as the xy-plane the moments of inertia will necessarily obey Jx + Jy = Jz and
a general inertia tensor can thus not be produced. ✷
4.1. THE ANGULAR MOMENTUM OF A RIGID BODY 63
Example 4.2 Calculate the inertia tensor with respect to a basis which has been rotated by
the angle ψ around e3 of the principal basis, using equation 4.11.
Solution: In the principal basis the matrix of the inertia tensor is given by equation 4.14.
The new matrix in the rotated system becomes
cos ψ − sin ψ 0 J x 0 0 cos ψ sin ψ 0
JO = ORA JA ARO = sin ψ cos ψ 0 0 Jy 0 − sin ψ cos ψ 0
0 0 1 0 0 Jz 0 0 1
cos2 ψJx + sin ψJy cos ψ sin ψ (Jx − Jy ) 0
2
This calculation thus shows that the product of inertia in the non-principal axis system is given
by
O
Dxy = − cos ψ sin ψ (Jx − Jy ). (4.31)
From this it is clearly seen that in the case of a symmetric top, Jx = Jy , the directions of the
X- and Y -axes, in the plane perpendicular to the Z-axis, do not matter since then Dxy ≡ 0.
When two principal moments of inertia are equal rotation about the third axis will not make
the inertia tensor matrix non-diagonal. ✷
In a principal axis system the rotational kinetic energy, equation 3.15, takes the
simple form
1
T = (Jx ωx2 + Jy ωy2 + Jz ωz2 ). (4.32)
2
The angular momentum vector components of equation 4.5 become simply
in a such a system.
where we have omitted the particle index, for simplicity, and where δkl = 1 when k = l
and δkl = 0 when k = l. If we define the spherical moment of inertia a by
JS ≡ mi ri · ri (4.35)
i
and then extend the definition of the products of inertia, 3.17, to diagonal terms so
that e.g.
Dxx ≡ mi xi xi , (4.36)
i
and similarly for y and z, we can rewrite the above formula for the inertia tensor matrix
in the form
1 0 0 Dxx Dxy Dxz
J = JS 1 − D = JS 0 1 0 − Dyx Dyy Dyz . (4.37)
0 0 1 Dzx Dzy Dzz
64 CHAPTER 4. DYNAMICS OF RIGID BODIES
We now see that the first term in this sum of matrices is invariant under the transfor-
mations 4.11. Eigen-values and eigen-vectors are thus completely determined by the
second part, the symmetric matrix D of products of inertia.
The division of the inertia tensor matrix into the two terms above simplify the
derivation of some of the properties of the inertia tensor. One such property is the
parallel axis theorem for the inertia tensor. We have come across the parallel axis
theorem for moments of inertia before. The corresponding theorem for the tensor is
found in exactly the same way. It states that the inertia tensor JC calculated with
respect to a point C of the body can be expressed in terms of the inertia tensor JG with
respect to the centre of mass G of the body and the vector from G to C: GC = R =
parallel axis (R1 , R2 , R3 ) according to the formula
theorem for
inertia tensor JC = JG + m(R2 1 − R ⊗ R). (4.38)
is in fact our old friend the parallel axis (or Steiner’s) theorem.
mr̈G = F (4.41)
where the first form is the general one while the second may be used when there is a
point C of the body which is fixed in space. We write simply
L̇ = M (4.43)
in what follows to simplify the notation. In order to take the time derivative of the
angular momentum vector, as given by equations 4.5, we must decide which basis
vectors the components refer to. In order for the inertia tensor matrix elements to be
guaranteed to be constant we must use a basis fixed in the body, i.e. a rotating basis.
The time derivative must thus be calculated using the formula 2.61, i.e.
Od Ad
L̇ = L= L + ω × L. (4.44)
dt dt
4.3. FIXED AXIS ROTATION AND REACTION FORCES 65
To simplify matters further we choose, among all possible basis vectors fixed in the
body, the principal basis, which we denote EA , and in which the angular momentum
vector has the simple form (see equation 4.33)
L = J1 ω1 eA A A
1 + J2 ω2 e2 + J3 ω3 e3 . (4.45)
We now find that
1 + J2 ω̇2 e2 + J3 ω̇3 e3 + ω × L.
L̇ = J1 ω̇1 eA A A
(4.46)
If we now expand the vector product in terms of components we find that the angular
momentum principle gives Euler’s dynamic
equations for the
J1 ω̇1 + (J3 − J2 )ω2 ω3 = M1 , rigid body
J2 ω̇2 + (J1 − J3 )ω3 ω1 = M2 , (4.47)
J3 ω̇3 + (J2 − J1 )ω1 ω2 = M3 .
These equations are called Euler’s dynamic equations. Together with Euler’s kinematic
equations, 2.69, which give the components of the angular velocity vector, in the same
body fixed frame, in terms of Euler angles, these equations determine the rotational
motion of the rigid body.
R Y
ϕ
G
X
m g
Figure 4.2: The physical pendulum is a rigid body which can rotate freely around the fixed
Z-axis. It is acted on by gravity mg at the centre of mass G. The reaction force R from the
axis is due to a smooth constraint and does no work.
Example 4.3 A straight homogeneous rod of length l and mass m has been suspended so
that it can swing freely about one of its ends.
a) Calculate the period for small plane oscillations around the vertical (equilibrium) position.
b) What length should a simple pendulum of the same mass have in order to have the same
period?
Solution:
a) In this case the moment of inertia is Jz = 13 ml2 and the distance from the rotation axis (at
the end of the rod) and its center of mass (in the middle) is h = l/2. Formula 4.52 then gives
# #
1 2
ml 2l
T = 2π 3 = 2π (4.53)
mgl/2 3g
Let us investigate how the small amplitude period changes when the position of the
axis is changed. According to the parallel axis theorem
where JzG is the moment of inertia with respect to a parallel Z-axis through the centre of
mass. Here dz and dGz are the radii of gyration corresponding to Jz and JzG respectively.
If we insert this into equation 4.52 we get
#
d2Gz + h2
Tp (h) = 2π . (4.55)
gh
This shows that both limh→0 Tp = ∞ and limh→∞ Tp = ∞. Consequently there must
be some value for h between zero and infinity for which the period is minimized. The
4.3. FIXED AXIS ROTATION AND REACTION FORCES 67
minimum h-value is easily found by putting the derivative of Tp with respect to h equal
to zero. This gives hmin = dGz and therefore
#
dGz
Tp (hmin = dGz ) = 2π 2 . (4.56)
g
is the smallest possible period for small amplitude oscillations of a physical pendulum.
The energy of the physical pendulum is given by
1
Jz ϕ̇2 − mgh cos ϕ = E. (4.57)
2
This equation can be used to find the angular velocity as a function of angle, ϕ̇(ϕ), as
in the following example.
Example 4.4 A straight homogeneous rod of length l and mass m has been suspended so
that it can swing freely about one of its ends. It is released with zero velocity in a horizontal
position. Calculate the angular velocity of the rod when it is vertical.
Solution:
In this case Jz = 13 ml2 and h = l/2. The initial conditions are ϕ(0) = π/2 and ϕ̇(0) = 0. We
put these into equation 4.57 and get E = 0. In the vertical position we then get, from the same
equation, that
1 1 2 l
ml ϕ̇ − mg
2
cos 0 = 0. (4.58)
2 3 2
Using cos 0 = 1 we easily solve this equation to get
3g
ϕ̇ = (4.59)
l
for the angular velocity of the rod in the vertical position. ✷
Example 4.5 A homogeneous circular cylinder of mass m and radius r rolls without slipping
on a cylindrical track of radius R(> r). See figure 4.3 for the geometry. Determine the period
for small oscillations of the cylinder around the equilibrium position.
Solution: Since there is rolling without slipping the mechanical energy is conserved and only
the gravitational force does work on the rolling cylinder so we can use the method of example
3.5. The kinetic energy can be found using the method of example 3.1.
In figure 4.3 two angles are introduced. The angle ψ is the angle that the vector from the
centre of the track to the midpoint of the cylinder makes with the vertical. The angle φ is the
angle between this vector and the line on the cylinder which is vertical at equilibrium. Because
of the rolling without slipping constraint we have the following connection between these:
Rψ = rφ. (4.60)
This says that the length Rψ along the track from the equilibrium position must be equal to
the length rφ along the circumference from the point of contact at equilibrium to the current
point of contact. Note, however, that the angle of rotation of the cylinder with respect to the
fixed vertical direction is given by φ − ψ.
The potential energy is given by Φ = mgh, where h is the height of the centre of mass
which is given by
h = (R − r)(1 − cos ψ). (4.61)
r
To get the total kinetic energy we need the angular velocity which is (note that ψ = R φ)
r r
ω = φ̇ − ψ̇ = φ̇ − φ̇ = (1 − )φ̇ (4.62)
R R
68 CHAPTER 4. DYNAMICS OF RIGID BODIES
r
R φ
Figure 4.3: This figure shows a non-equilibrium position for a cylinder rolling on a cylindrical
track. In example 4.5 the period for small oscillations of the rolling cylinder around the equi-
librium at the bottom of the track, is calculated. The angle φ is the angle that the cylinder has
turned when the vector from the center of the track to its mid-point has rotated the angle ψ.
so T is given by
1 1 1 r
T = JC ω 2 = (mr2 + mr2 )(1 − )2 φ̇2 . (4.63)
2 2 2 R
r
Using ψ = Rφ again we now get for the total energy
1 3 2 r r
Em = ( mr )(1 − )2 φ̇2 + mg(R − r)[1 − cos( φ)]. (4.64)
2 2 R R
We can now differentiate this with respect to time. When the resulting expression is divided by
φ̇ we get an equation of motion. In this equation one then makes use of the assumption of small
amplitude and put sin( Rr φ) ≈ Rr φ. One then gets the usual equation of motion for harmonic
oscillations and the period becomes
#
6(R − r)
Tp = π . (4.65)
g
F = mr̈G (4.66)
and here F = Fa + R where Fa is the ‘applied’ force. In figure 4.2 the only applied
force is gravity so Fa = mg but in the general case there may be other known external
forces applied to the body. If we now write down the three Cartesian components of
the equation of motion above we get
In order to make use of the known rotational motion it is, however, better to introduce
cylindrical coordinates and take the components along the position dependent basis
vectors eρ and eϕ . The acceleration of the centre of mass, for which ρ = h, is then
given by
r̈G = −hϕ̇2 eρ + hϕ̈ eϕ . (4.70)
The ρ- and ϕ-components of the equations of motion above now give
These equations thus determine the interesting components of the reaction force on
the body from the axis. The force −R is then, of course, the force on the axis and its
bearings from the body.
We now calculate the right hand sides as explicitly as possible for the case of figure
4.2. We have that
Figure 4.4: A rigid body can rotate around a fixed Z-axis. The axis is held in place by
bearings at A and at O. The reaction forces at these are R1 and R2 respectively. There is also
an applied force Fa , normally its weight, acting on the body.
There are, however, now four unknowns but still only two equations, so this does not
suffice to determine the separate bearing reactions.
We’ll now show that the missing equations are provided by the two unused x- and
y-components of the principle of angular momentum L̇O = MO . Since the angular
velocity vector is ω = ϕ̇ ez equations 4.5 give us (as usual we leave out the base point
index O)
To find the time derivative L̇ we use equation 4.44 and find for the x- and y-components
d
L̇x = (−Dxz ϕ̇) + ωy Lz − ωz Ly = −Dxz ϕ̈ + Dyz ϕ̇2 , (4.80)
dt
d
L̇y = (−Dyz ϕ̇) + ωz Lx − ωx Lz = −Dyz ϕ̈ − Dxz ϕ̇2 . (4.81)
dt
We now need the corresponding components of the moment. Reference to figure 4.4
immediately gives (zA is the distance between the two bearings)
so the angular momentum principle now provides us with the two equations for the
reaction force at bearing A:
1
R1y = (Dxz ϕ̈ − Dyz ϕ̇2 + Mxa ), (4.84)
zA
1
R1x = − (Dyz ϕ̈ + Dxz ϕ̇2 + Mya ). (4.85)
zA
These two equations together with the equations 4.78 completely solve the problem i.e.
determine the four components, perpendicular to the axis, of the the two reaction forces
4.4. PROBLEMS 71
A O
ϕ N
C
M
T
x
B
A B
Figure 4.5: The figure on the left refers to problem 4.1. The notation used in the problem is
indicated. The rod can swing around a horizontal axis through A.
Figure 4.6: The figure on the right refers to problem 4.2. The straight narrow rod, which
hangs in two strings of equal length, one from each end, can swing as a pendulum in a vertical
plane.
at the bearings. The force along the axis can be determined if it is assumed to arise
from one of the bearings, in which case it is simply the negative of the z-component
of the applied force. Should both bearings apply forces along the axis the problem is
statically indeterminate.
If one wishes to minimize the reaction forces on the bearings it is obvious from
equations 4.71 that one should make h = 0, i.e. put the centre of mass on the axis,
since then only the applied force will contribute to the net reaction force R. A rotating
system for which this has been done is sometimes called ‘statically balanced’. If the
total force is made small in this way there is, however, still no guarantee that the
individual forces on the two bearings of this section are small. To achieve this one must
also make the products of inertia Dxz and Dyz zero, according to our findings above.
This means that the rotation axis should be a principal axis of the body. The system
is then said to be ‘dynamically balanced’. Wheels that rotate rapidly should thus be
both statically and dynamically balanced. An example of a system which is statically
but not dynamically balanced is the one of example 1.1 on page 6.
4.4 Problems
Problem 4.1 A straight narrow homogeneous AB rod of mass m and length # can
swing, with negligible friction, around a fixed horizontal axis at A, as shown in figure
4.5. The instantaneous angle that it makes with the vertical is ϕ and the maximum
value for this angle is α. Calculate, as functions of ϕ and x, the tension N , the shear
T , and the bending moment M on the part CB of the rod in an imaginary cut at C a
distance x from B.
Problem 4.2 A straight narrow rod AB, hangs in two strings of equal length, one
from each end and both fixed at O. The rod can swing as a pendulum in a vertical
plane around a horizontal axis through O, see figure 4.6. The lengths of the rod and
the strings have the ratio 6/5(= |AB|/|AO|). What maximum value of the angle of
deflection is allowed if the strings are to remain taut at all times?
72 CHAPTER 4. DYNAMICS OF RIGID BODIES
2a
2a
A B
k
C
R
Figure 4.7: The figure on the left refers to problem 4.3. A circular disc hangs in a string fixed
at the two points A and B. The string goes round the circumference of the disc. The part BC
of the string has been replaced by a spring.
Figure 4.8: The figure on the right refers to problem 4.4. A cube stands on a semi-cylinder.
Problem 4.3 A circular homogeneous circular disc of mass m and radius R hangs
vertically in a string fixed at the two points A and B of a horizontal ceiling. The points
are a distance 2R apart. The string goes round the circumference of the disc one and
a half times. The part BC of the string has been replaced by a spring of stiffness k, as
shown in figure 4.7. The string will not slip on the circumference of the disc. Calculate
the period for small vertical oscillations.
Problem 4.5 A tall factory chimney made of brick is being demolished. It is severed
at the base by means of an explosion an starts to fall. Assume constant thickness.
Assume that it breaks again somewhere during the fall. Use the results of problem 4.1
to calculate where this is likely to happen and in which direction.
4.5. HINTS AND ANSWERS 73
l + 2x
R x=Rϕ
Figure 4.9: This figure refers to answer 4.3. It shows why the elongation of the spring is twice
the vertical displacement, x, of the disc.
Answer 4.2 Do the experiment at home and observe on which side the string first
becomes slack. The maximum allowed angle is given by
19
α = arctan
4
and this is the answer.
Answer 4.3 The string has fixed length and must go one and a half times round the
disc at all times. The disc then translates and rotates in such a way that the vertical
displacement is R times the rotation angle, see figure 4.9. The elongation of the spring
must be twice the vertical displacement of the disc. One finds that
3m
T =π
2k
is the period of the oscillations.
74 CHAPTER 4. DYNAMICS OF RIGID BODIES
Answer 4.5 The chimney is likely to break where the bending moment M (x, ϕ) has
its maximum. This happens at x = 2#/3.
Chapter 5
In this chapter we discuss the general non-planar rotational motion of rigid bodies.
In particular we discuss the free motion (no external moment) and the motion of the
symmetric top in the gravitational field.
The general rotational motion of an asymmetric top rigid body is in principle given
by solving Euler’s dynamic equations 4.47 and Euler’s kinematic equations 2.69 together
as a system of coupled non-linear differential equations for the three functions ψ(t), θ(t),
and ϕ(t). In the general case, for some given external moment M acting on the body,
the only method for finding a solution is by numerical techniques. In the free case
when the external moment is zero M = 0 it is possible to find a solution of Euler’s
dynamic equations in terms of elliptic functions but we will not go into this. In order
to get reasonably simple closed form solutions we concentrate on the symmetric top
with principal moments of inertia given by J1 = J2 = J3 . For the free asymmetric top
we discuss the Poinsot construction and stability of rotation around the principal axes.
Example 5.1 A sphere of mass m and radius R rolls on a rough inclined plane. The angle
of inclination is β, see figure 5.1. The moment of inertia of the sphere is J for any axis through
its center of mass (=centroid) G. Let the X-axis point downwards along the incline.
a) Find the trajectory of the sphere on the plane for arbitrary initial conditions.
b) Use conservation of mechanical energy to find an equation connecting x and ẋ.
Solution:
a) Since the sphere rolls without slipping the point C on the sphere in contact with the plane is
75
76 CHAPTER 5. THREE DIMENSIONAL MOTION OF RIGID BODIES
Figure 5.1: The notation used in example 5.1 is defined in this figure. The sphere rolls without
slipping on the inclined plane.
instantaneously at rest. The motion of the sphere is thus pure rotation around C. The moment
of inertia of the sphere with respect to any axis in the plane through C is, according to the
parallel axes theorem,
JC = J + mR2 . (5.2)
We will use the equation of motion L̇C = MC and we thus need expressions for LC and MC .
These are found to be (see equation 1.12)
and
MC = M = CG × W = Rmg sin β ey (5.4)
where W = −mg(cos β ez − sin β ex ) is the weight of the sphere as indicated in figure 5.1. Since
vC = 0 the connection formula for velocities in a rigid body, equation 2.74, gives us
vG = v = ω × CG = (ωx ex + ωy ey + ωz ez ) × R ez (5.5)
for the center of mass velocity, v, of the sphere. The three components of this equation can be
written
ẋ = Rωy (5.6)
ẏ = −Rωx (5.7)
ż = 0 (5.8)
If we put
Rmg sin β
α ≡ ω̇y = (5.15)
J + mR2
for the angular acceleration around the Y -axis, we can solve these equation for the angular
velocity components as follows
This can be inserted into equation 5.6 above and integrated to give
Note that gravity is the only force doing work on the sphere since the other forces, normal force
and (static) friction, do no work. Thus the mechanical energy
E =T +Φ (5.26)
is conserved. Since ωx and ωz are also constants use of equation 5.6 allows us to write
2
1 1 1 ẋ
const. = E + (J + mR2 )ωx2 + Jωz2 = (J + mR2 ) − mgx sin β. (5.27)
2 2 2 R
This concludes our treatment of the sphere rolling down a rough incline. ✷
The angular velocity of the frame B relative to the fixed frame O is given by
O
ω B = ψ̇ eO3 + θ̇ eB1 = θ̇ eB1 + ψ̇(sin θ eB2 + cos θ eB3 ). (5.29)
78 CHAPTER 5. THREE DIMENSIONAL MOTION OF RIGID BODIES
Figure 5.2: To study the motion of the symmetric top there is no need to use the body fixed
basis EA . Because of the equality of two of the moments of inertia all elements of the inertia
tensor matrix are constant already in the basis EB where the 1,2-plane corresponds to the plane
of the equal moments of inertia. For the free symmetric top the origin O is taken at the center
of mass (O = G) and the third axis of the observer fixed system, eO3 , is taken along the constant
angular momentum vector L.
According to the theorem on the additivity of angular velocities the angular velocity
of the body relative to the fixed reference frame is then
(compare equation 2.65 where the same vector is given in the A basis).
We select the fixed 3-axis along the (constant) angular momentum vector so that L =
L eO
3 . The components of L in the B basis are then given by
0 = J1 θ̇,
L sin θ = J1 ψ̇ sin θ, (5.33)
L cos θ = J3 (ψ̇ cos θ + ϕ̇).
5.2. THE SYMMETRIC TOP 79
The first of these equations says that θ = θ0 =const. I.e. the axis of the top makes a
constant angle with the angular momentum vector L. The second of the three equations
gives us
L
ψ̇ = (5.34)
J1
unless θ0 = 0. This is the angular velocity with which the axis of the top rotates around
L. The third equation gives
L J1 − J3
ϕ̇ = cos θ0 (5.35)
J1 J3
for the angular velocity of the body in the B frame. Note that the sign of this angular
velocity depends on the shape of the body; it is positive when J1 > J3 as is the case for
the ‘prolate’ body in figure 5.2 (assuming it is homogeneous) but it would be negative
for an ‘oblate’ (flattened) body.
In the case θ0 = 0, when the axis of the top is parallel to the angular momentum,
the third of our equations give L/J3 = ψ̇ + ϕ̇ and this is simply the magnitude of the
angular velocity of the top around the common fixed direction of L and its axis. In this
case one cannot resolve the two angular velocities since the B frame becomes undefined.
This last result, that if the symmetric top spins around its axis, then this axis is
parallel to the angular momentum vector and is fixed in space, is the principle behind
the gyroscope. A symmetric top body which has been mounted in a pair of gimbal
rings so that the external moment on it is negligible will be a free symmetric top. If
it spins around its symmetry axis this axis will have a fixed direction in space so the
system can be used as a compass.
Example 5.2 Consider Euler’s dynamic equations 4.47 for the free symmetric top:
Here ωi , (i = 1, 2, 3) are the body fixed components of the angular velocity vector. Find the
time dependence of these.
Solution:
The third of the equations 5.36 immediately gives
where
J3 − J1
β≡ ω3 (0). (5.40)
J1
If we now multiply the equation for ω2 by the imaginary unit i and add the result to the
equation for ω1 we find that the quantity
ζ ≡ ω1 + iω2 (5.41)
obeys
ζ̇ − iβζ = 0. (5.42)
The solution to this simple differential equation is
Figure 5.3: Just as the free symmetric top the heavy symmetric top is best studied in the
basis EB . The origin O is taken at the (fixed) point of contact, C, with the ground (O = C)
and the third axis of the observer fixed system, eO
3 , is taken along the vertical direction. The
moment is given by M = CG × mg = −# eB 3 × mg eO B
3 = mg# sin θ e1 .
where A is a complex constant that depends on the initial conditions. The result is thus that
where α is the argument of A = |A| exp(iα). This means that the angular velocity vector, in
the body fixed system, rotates with angular velocity β, around the third principal axis (the one
with different moment of inertia). ✷
Here m is the mass of the body and # is the distance between G and C. We have
used formula 4.38 in which we have put that the components of R in the B coordinate
system are (R1 R2 R3 ) = (0 0 #). Note that this inertia tensor in the B coordinate
system is exactly the same as in the body fixed A system because of the symmetry
(i.e. equality of the principal moments of inertia), just as we found in our treatment of
the free symmetric top. The reason, in both cases, being that the body fixed frame is
obtained by a rotation of the B frame an angle ϕ around the axis of the top.
The angular velocity vectors of the frame B and the body are given by the same
expressions as in equation 5.30 and those above it. We now calculate the equations of
motion L̇ = M and we wish the components in the B frame. We get
BdL
L̇ = + ωB × L
O
(5.47)
dt
according to equation 2.61 and we have that
L = Jˆω = J(
ˆ Oω B + B
ω A ) = J1 (θ̇ eB1 + ψ̇ sin θ eB2 ) + J3 (ψ̇ cos θ + ϕ̇)eB3 . (5.48)
Here we have introduced J1 ≡ J1 + m#2 . If we now use the expression in equation 5.29
for Oω B we can calculate the vector product that we need for the time derivative of L
in the B frame. The moment of the weight of the body can be read off directly from
figure 5.3 and is
M = mg# sin θ eB1. (5.49)
Finally the angular momentum law, L̇ = M, now gives us
d
(J θ̇) + [(J3 − J1 )ψ̇ 2 sin θ cos θ + J3 ψ̇ ϕ̇ sin θ] = mg# sin θ (5.50)
dt 1
d
(J ψ̇ sin θ) − [(J3 − J1 )ψ̇ θ̇ cos θ + J3 θ̇ϕ̇] = 0 (5.51)
dt 1
d
[J3 (ψ̇ cos θ + ϕ̇)] = 0 (5.52)
dt
Of these complicated equations only the third gives some simply useful information. It
tells us that
L3 ≡ J3 (ψ̇ cos θ + ϕ̇) = const. (5.53)
i.e. that the component of the angular momentum along the axis of the top is conserved.
The reason for this is that neither the vector product Oω B × L nor the moment M
have components along eB 3 , and this depends crucially on the fact that the two first
moments of inertia are equal. The corresponding conserved quantity therefore does not
exist for an asymmetric top.
Since the moment M always lies in the horizontal plane, M = Mx eO O
1 + My e2 , one
can conclude that the vertical component of L also must be conserved; just consider
the third (Z-component) of L̇ = M in the fixed O-system. Use of equation 5.48 gives
us the following expression for this component of the angular momentum vector
L · eO
3 = L · (sin θ e2 + cos θ e3 ) = ψ̇(J1 sin θ + J3 cos θ) + J3 ϕ̇ cos θ.
B B 2 2
(5.54)
for the heavy symmetric top. The constraint force acting on the top at C does no work
so the only work done on the body is done by the conservative force of gravity. This
means that the mechanical energy of the top is conserved. We’ll use these conserved
quantities later to study the ‘nutation’.
82 CHAPTER 5. THREE DIMENSIONAL MOTION OF RIGID BODIES
Ω = ψ̇ = const., (5.56)
θ0 = θ = const. (5.57)
for the constant angular velocity of precession and the constant angle between axis and
vertical respectively. When this is inserted in equation 5.53 one immediately finds that
also the angular velocity of the top around its own axis must be constant
L3
ϕ̇ = − Ω cos θ0 = const. (5.58)
J3
The third equation 5.52 is then automatically satisfied. The second equation, 5.51, is
easily seen to satisfied by this type of motion; all its terms become zero. The first
equation (5.50) gives after some algebraic manipulation that
1 mg#
ϕ̇ = − Ω(J3 − J1 ) cos θ0 . (5.59)
J3 Ω
mg#
L3 = + J1 Ω cos θ0 . (5.60)
Ω
all these results agree and constitute one possible solution of the equations of motion.
One finds the following expression for the angular velocity of precession
mg#
Ω= . (5.61)
J3 ϕ̇ − (J3 − J1 )(ϕ̇ − L3 /J3 )
One might call this kind of motion ‘pure’ precession. It is characterized by the constancy
of the angle θ. Changes in θ are called nutations and are treated below.
It is fairly easy to see that the precessional motion always is a good approximation
to the exact motion when the top spins rapidly enough. If ϕ̇ is very large one should
have L ≈ J3 ϕ̇ eB 3 . Then however L is parallel to the moment arm of the weight and we
get from L̇ = M that (in the fixed reference frame)
OdL m#
+ g×L=0 (5.62)
dt L
after moving the moment to the left hand side. This, however, is the equation (see
equation 2.61)
PdL OdL
= + Pω O × L = 0. (5.63)
dt dt
that L would obey if it was at rest in a reference frame P with respect to which O
rotates with the angular velocity
m# m#g O
P
ωO = g=− e . (5.64)
L J3 ϕ̇ 3
5.2. THE SYMMETRIC TOP 83
The conclusion must be that L and with it the axis of the top, rotates around the
vertical axis with an angular velocity Oω P which is the negative of Pω O . This angular
velocity is thus
m#g O
Ω= e . (5.65)
J3 ϕ̇ 3
If we compare with the result 5.61 above we see that these agree reasonably with each
other when ϕ̇ ≈ L3 /J3 and according to equation 5.58 this happens when |Ω cos θ0 |
|L3 /J3 |. It is thus exact also when θ0 = π/2.
1 L2
E = J1 (θ̇2 + ψ̇ 2 sin2 θ) + 3 + mg# cos θ (5.66)
2 2J3
where the constant L3 is given in equation 5.53. However, using the two conserved
components of the angular momentum in equations 5.53 and 5.55 one can easily find
that
Lz − L3 cos θ
ψ̇ = (5.67)
J1 sin2 θ
and thus also eliminate ψ̇ from the expression for the energy. The result is that
1
E = J1 θ̇2 + Φeff (θ) (5.68)
2
where E is the constant
L23
E ≡ E − − mg# (5.69)
2J3
and
(Lz − L3 cos θ)2
Φeff (θ) ≡ − mg#(1 − cos θ). (5.70)
2J1 sin2 θ
Conservation of energy has thus given us a one-dimensional effective energy conserva-
tion law for the θ-motion.
We can now find the limits (turning points) of the θ-motion as the roots of the
equation
E − Φeff (θ) = 0. (5.71)
It turns out that there are in general two roots, θ1 ≤ θ2 , of this equation between which
‘nutation’ takes place. When the two roots coincide one has the special case of pure
precession.
Example 5.3 Find the condition for the rotation of the heavy symmetrical top in a vertical
position to be stable.
Solution: For θ = 0 the two vectors eB O
3 and e3 coincide so we must have that L3 = Lz .
We also see that E = 0. The rotation will be stable if the function Φeff (θ) has a minimum at
θ = 0. To check if this is the case one puts L3 = Lz in and then approximate the trigonometric
functions according to sin θ ≈ θ and cos θ ≈ 1 − 12 θ2 . The result is that
L23 mg#
Φeff (θ) ≈ − θ2 . (5.72)
8J1 2
This function thus has a minimum when L23 > 4J1 mg#. So when the angular momentum is
large enough this motion is stable, but when friction has reduced the angular momentum to
this limiting value the top starts to wobble and soon falls to the ground. ✷
84 CHAPTER 5. THREE DIMENSIONAL MOTION OF RIGID BODIES
ω = Ω + ϕ̇ eB3 (5.73)
(J1 Ω̇1 , J1 Ω̇2 , J3 (Ω̇3 + ϕ̈)) + Ω × (J1 Ω1 , J1 Ω2 , J3 (Ω3 + ϕ̇)) = (M1 , M2 , M3 ). (5.76)
and they are clearly linear in M(t) and ϕ(t). If ϕ(t) is known one can calculate the
moment M, or if one component of M is known one can calculate ϕ(t) and the other
two components of M.
Example 5.4 On a light horizontal axis OB, of length #, is mounted a bicycle wheel of mass
m that can rotate freely around it at B, see figure 5.4. The moment of inertia of the wheel,
which can be considered to be thin, with respect to the axis OB is J and the axis is connected
by a ball and socket joint at O to a fixed vertical axis. This vertical axis rotates with constant
angular velocity Ω. By means of an arm, which has a smooth ring at A through which OB goes,
it keeps the axis OB horizontal, and carries it around in the rotation. The angular velocity of
the bicycle wheel around the axis OB is ϕ̇. Calculate the force F at A from the ring on the
axis OB, assuming that the distance |OA| = d (< #).
Solution:
The symmetric top in this case is the bicycle wheel together with the axis OB which also is the
symmetry axis. We introduce a B-system just as in figure 5.3 so that eB 3 is along OB, the angle
θ = π2 and Ω = ψ̇. The angular velocity vector of the B-system is thus
Ω = Ω1 e B B B B
1 + Ω 2 e 2 + Ω3 e 3 = Ω e 2 (5.78)
1
The name refers to the french nineteenth century mathematician H. Resal.
5.2. THE SYMMETRIC TOP 85
Ω B
e2
O B
. eB3
A ϕ
Figure 5.4: This figure shows the rotating bicycle wheel discussed in example 5.4. The vertical
axis through O together with its arm to A rotates with angular velocity Ω around the vertical
direction. The bicycle wheel rotates with angular velocity ϕ̇ on the light horizontal axis OB.
so that Ω1 = Ω3 = 0, Ω2 = Ω while all time derivatives are zero: Ω̇1 = Ω̇2 = Ω̇3 = 0. Only the
third term in the first two of equations 5.77 will thus survive. Since the wheel is thin one must
have J1 = J2 = J/2 for axes through B, in the plane of the wheel. The fixed base point is at
O, however, so we really need J1 = J1 + m#2 . In this case therefore the quantity (J3 − J1 ) of
equations 5.77 is given by J − (J/2 + m#2 ) = J/2 − m#2 . However, we found above that we
don’t need this quantitiy, only J3 = J is needed. The system of equations 5.77 gives
JΩϕ̇ = M1 ,
0 = M2 , (5.79)
J ϕ̈ = M3 .
1 + M2 e2 + M3 e3 = OB × (−mg e2 ) + OA × F,
M = M1 e B B B B
(5.80)
where F is the unknown force acting at A. Since the ring is smooth this force cannot have any
component along OB and can thus be expressed in the form
F = F 1 eB B
1 + F2 e 2 . (5.81)
Since OB = # eB
3 and OA = d e3 we now find that
B
M = (mg# − F2 d) eB B
1 + F1 d e 2 . (5.82)
JΩϕ̇ = mg# − F2 d,
0 = F1 d, (5.83)
J ϕ̈ = 0,
Note that this equation defines a quadratic surface in ‘angular velocity space’. Since the
inertia tensor is positive definite this surface is an ellipsoid which is called the inertia
ellipsoid. Let us denote the components of the angular velocity, and the inertia tensor
with respect to a space fixed basis by ωα and Jαβ respectively, where α, β = x, y, z. If
we now put
f (ω , t) ≡ Jαβ (t) ωα ωβ , (5.86)
αβ
where we we have indicated explicitly the time dependence of the elements of the inertia
tensor (due to the rotation of the body), we can write the conservation of energy
f (ω , t) = 2E = const. (5.87)
For each value of the time this expression represents a quadratic surface in angular
velocity space, namely the inertia ellipsoid rotating with the body.
where the symmetry of the inertia tensor has been used. If we compare with equations
4.5 we find that these components are, in fact, two times the components of the angular
momentum
∂f
= 2Lα . (5.89)
∂ωα
But the space fixed components of the (conserved) angular momentum L must be
constants, so this shows that the time dependence of the inertia tensor components are
such that they cancel the time dependence of the angular velocity components in the
expression Lα = β Jαβ (t)ωβ (t) = const.
The energy equation can also be written
2E = L · ω = Lα ωα (t). (5.90)
α
This is the equation for a (fixed) plane in angular velocity space, with L as normal
vector. This plane is called the invariable plane. Since L is perpendicular to the inertia
ellipsoid, at the point ω (t), the invariable plane must be tangent to the ellipsoid at this
point, see figure 5.5. The angular velocity vector, which can be regarded as defining
the instantaneous axis of rotation of the body (passing through its center of mass), is
thus seen to move in such a way that its tip always is both on the inertia ellipsoid and
on the invariable plane. These surfaces are thus always in contact and the tip of the
angular velocity vector traces out a curve on each surface, the two curves touching at
5.3. THE FREE ASYMMETRIC TOP 87
Figure 5.5: This figure illustrates the Poinsot construction. The inertia ellipsoid is in contact
with the invariable plane which has L as fixed normal vector.
the point of contact. The curve on the inertia ellipsoid is called the polehode and the
curve on the invariable plane is called the herpolehode.
In summary we have thus found that the motion of the free asymmetric top can be
described as the rolling (without slipping) of the inertia ellipsoid on the invariable plane.
The angular velocity vector goes from the center of the inertia ellipsoid (the origin and
the center of the body) to the instantaneous point of contact. This geometric view of
the dynamics was first given by Poinsot in 1834.
Using these we can check whether there are solutions of the form
where i = 1, 2, or 3. When this ‘ansatz’ is inserted into the equations above two of
them give 0 = 0 and the third, the i-equation, gives
Ji ω̇i = 0. (5.93)
This equation is satisfied if ωi =constant and we have thus found that rotation around
one of the principal axes, with constant angular velocity, is one possible motion of the
free top. For this simple rotation the angular velocity vector is parallel to the angular
momentum, L = Ji ωi eA i . Consequently this principal axis then has a fixed direction
in space.
88 CHAPTER 5. THREE DIMENSIONAL MOTION OF RIGID BODIES
In practice there will always be small perturbations on a body and one might ask
whether the principal axis parallel rotations, found above, are stable. For the motion
of satellites in space, which may be required to point in some fixed direction, this is
a question of practical interest. To investigate this stability we rewrite the system of
equations 5.91 in the form
ω̇i = Ci ωj ωk , i, j, k = 1 → 2 → 3 → 1. (5.94)
J3 − J2 J3 − J1 J2 − J1
C1 ≡ < 0, C2 ≡ > 0, C3 ≡ < 0. (5.96)
J1 J2 J3
We now assume a perturbed rotation around the j-axis: ω (t) = ωj0 eA j + δ ω (t). Here
δ ω is a small perturbation to the exact unperturbed solution. The components of this
ω (t) are
ωi (t) = ωj0 δji + δωi (t) (5.97)
where δij is the Kronecker delta (see equation 2.12) and summation over j from 1 to 3
is implied. When this is inserted into the equations 5.94 one gets
If we differentiate the second of these equations with respect to time and insert the
third equation in the resulting expression we get
If we neglect higher powers of the perturbations and their time derivatives this gives
¨ i ≈ Ci Ck (ω 0 )2 δωi .
δω (5.102)
j
This equation shows that the perturbation will have an oscillatory behavior provided
Ci Ck < 0, otherwise it will be exponential. For j = 1 this constant is C2 C3 and
negative, for j = 3 it is C1 C2 and also negative, but for j = 2 the constant is C1 C3 and
consequently positive. This shows that rotation around a principal axis with constant
angular velocity is stable for the axes corresponding to the maximum and minimum
moments of inertia but unstable when the axis corresponds to the middle moment of
inertia.
5.4 Problems
Problem 5.1 The following problem illustrates the behavior of a bowling ball. A
homogeneous solid sphere of radius R is, at time t = 0, given a center of mass velocity
v(0) = v0 ex and an angular velocity ω (0) = ω0 ex and is placed on a rough horizontal
floor that coincides with the xy-plane of the coordinate system. Calculate the center
of mass velocity v of the sphere when is has started to roll without sliding.
5.4. PROBLEMS 89
a a
Ω
l
a
Figure 5.6: This figure refers to problem 5.3 and shows the wheel, of radius a, from the edge.
Problem 5.2 An astronaut has managed to set a thin flat homogeneous metal square
spinning freely in space with no translational velocity. The component of the angular
velocity around the axis through the center of the square and perpendicular to it, is ϕ̇.
This axis precesses (rotates) around a fixed direction in space with which it makes an
angle θ = π/4. What is the angular velocity of this precession?
Problem 5.3 A wheel in the shape of a homogeneous circular disc of mass m and
radius a can rotate around a light axis of length l. The other end of this axis is
attached to a vertical axis by a hinge, so that it can rotate freely about a horizontal
axis, see figure 5.6, but when the vertical axis rotates the axis of the wheel must follow
and have the same vertical component of angular velocity. The hinge on the vertical
axis is at a distance a below a rough horizontal plane.
Assume that the vertical axis rotates with angular velocity Ω and that the wheel
rolls without slipping on the rough plane above it. Clearly, if Ω is too small this cannot
happen, and if Ω is zero the wheel and its axis will hang straight down. What is the
minimum value of Ω needed for this rolling to take place?
Problem 5.4 A (millstone) wheel of radius r and mass m is mounted on a light axis
OG as shown in figure 5.7. The axis is connected to a fixed ball and socket joint at O
and rotates around a vertical axis with constant angular velocity Ω. The angle between
the axis OG and the vertical has the constant value β = π/3. The wheel, which has
moment of inertia J3 with respect to the axis OG and J = J1 = J2 for all perpendicular
axes through O, rolls without slipping on the horizontal ground so that the geometric
contact point traces out a circle of radius r.
a) Find the magnitude, ω, of the angular velocity vector ω of the wheel.
b) Find the moment M = MO acting on the wheel.
c) Calculate the force N acting on the wheel from the ground at the point A in terms
of Ω, β, and mg.
90 CHAPTER 5. THREE DIMENSIONAL MOTION OF RIGID BODIES
r
mg
N
r
A
Figure 5.7: The figure refers to problem 5.4 and shows the wheel, of radius r, head on. It also
indicates the geometry of the problem. Note that the point of contact of the wheel with the
ground is at a distance r from the point on the ground below the ball and socket joint at O.
mv̇ = F,
J ω̇ = −R ez × F,
vC = v + R ez × ω ,
where J = 2mR2 /5 is the moment of inertia of the solid sphere, and F is the friction
force from the floor. Eliminating F by putting the first equation into the second we get
J ω̇ = −R ez × mv̇
J ω̇x = Rmÿ,
J ω̇y = −Rmẍ,
J ω̇z = 0.
ẋ = ẋC + Rωy ,
ẏ = ẏC − Rωx ,
ż = żC .
5.5. HINTS AND ANSWERS 91
Answer 5.3 Use the methods of example 5.4. The rolling condition is
lΩ = aω
where N is the normal force from the plane on the wheel. The angular velocity of the
B-system is, just as in example 5.4, given by
Ω = Ω eB
2.
92 CHAPTER 5. THREE DIMENSIONAL MOTION OF RIGID BODIES
Here J = 12 ma2 is the moment of inertia of the wheel with respect to its axis. The
condition that the normal force is positive (N > 0), and the rolling condition (lΩ = aω)
now give
l 1
0 < lN = Ω Ω ma − lmg.
2
a 2
Thus one finds that
2g
Ω>
a
is the answer.
Answer $5.4 √
a) ω = 2(1 + cos β)Ω = 3Ω,
b) Using the standard basis of the (B) Resal system (see figure 5.3) we find that
M = [J3 (1 + cos β) − J cos β]Ω2 sin β eB
1.
c) N = mg(1 + cos β) + r [J3 (1 + cos β) − J cos β]Ω2 sin β.
1
Chapter 6
Impact
Impact is characterized by the occurrence of large forces acting for short times and is the
technical term used to describe things like collisions, bounces and similar phenomena
where there are large, rapid velocity changes. As an idealized limit one can then use
the approximation that the time is zero and the force infinite so that a finite impulse
results. The velocities of particles and rigid bodies then change instantaneously with no
change in position. Momentum and angular momentum are useful concepts in dealing
with impact phenomena. Energy, on the other hand, is normally not conserved since
the large forces involved usually produce sound, heat and irreversible deformation.
If there is also an ordinary force, Fa , acting, the total impulse, or change of momentum,
is
ti +τ
∆p = p(tf ) − p(ti ) = F(t ) dt = (6.3)
ti
ti +τ
[Fa (t ) + K(t )] dt ≈ Fa (ti ) τ + S. (6.4)
ti
Since we assume that the force Fa remains of normal magnitude during the impact the
limit τ → 0 gives
p(tf ) − p(ti ) = S (6.5)
We can thus say that the the words ‘large’ for K and ‘short’ for τ apply when this
approximation, the ‘impulse approximation’, is valid.
93
94 CHAPTER 6. IMPACT
S
G S G2 G
1 en
1 en G
2
Figure 6.1: This figure illustrates some concepts presented in the text. It shows the point of
impact S and the normal of impact en . The impact to the left is central and that to the right
is eccentric.
in the limit when τ → 0. With suitable notation we can rewrite this result as follows
i.e. the total linear momentum just before the impact is the same just after the impact.
The corresponding calculation based on the angular momentum principle,
d
(L1 + L2 ) = Me , (6.17)
dt
gives in the same way
L1f + L2f = L1i + L2i . (6.18)
Here any continuously moving base point may be used since the extra term arising from
the velocity of the base point, see equation 1.50, gives zero contribution when τ → 0.
Equation 6.5 can be written for body 1, on the form
where S12 is the impulse of the impact force from body 2 on body 1. According to the
law of action and reaction (or Newton’s third) one has that S12 = −S21 . For body 1
we also have
L̇1 = Ma1 + rS × K12 (6.20)
where the origin is base point. Time integration of this from ti to ti + τ gives
when τ → 0 if we take into consideration the fact that rS , the position vector of the
point of impact, does not move during the impact. There is, of course, a corresponding
equation for body 2. If one chooses the point of impact S as origin (base point) this
equation gives the simple result
LS1f = LS1i . (6.22)
Provided that one knows the impulse of impact S12 one can use the three formulae
above to get the velocity state of the body just after impact in terms of the state just
before impact.
Example 6.2 A bat in the form of a straight narrow homogeneous rod of mass m and length
# is hit by a ball when at rest. The ball imparts an impulse S to the bat perpendicular to it at
the distance x from one of its ends, see figure 6.2.
a) Calculate the kinetic energy T (x) of the bat after the hit.
b) Is there a point on the bat which is at rest immediately after the impact? If so, where is it?
c) Which values of x correspond to maximal speed of the point of impact after the hit?
Solution:
96 CHAPTER 6. IMPACT
S
x
Figure 6.2: A bat in the form of a narrow homogeneous rod receives an impulse S at the
distance x from one end. The properties of the subsequent motion are discussed in example
6.2.
a) To find the kinetic energy we need to know the translational and angular velocities after the
impact. Use of equations 6.19 and 6.21 give us directly
S ey = mvG , (6.23)
#
(x − ) ex × S ey = JG ϕ̇ ez , (6.24)
2
where JG = 121
m#2 . Thus vG = S/m and ϕ̇ = 12S(x − #/2)/(m#2 ). The kinetic energy can now
be calculated from
1 1
T = mvG2 + JG ϕ̇2 (6.25)
2 2
and we get
2 2 2
1 S 1 1 12S(x − #/2) S2 x 1
T (x) = m + m#2
= 1 + 12 − . (6.26)
2 m 2 12 m#2 2m # 2
This shows directly that the kinetic energy has a minimum for x = #/2. This happens when the
ball hits in the middle of the rod so that only translational and no rotational motion results.
b) Equation 2.95 gives the x-coordinate (and the y-coordinate) of a point rigidly connected to
the body which has zero velocity. If we take the point A to be the centre of mass of the rod we
get
ẏG # S/m
xC = xG − = − . (6.27)
ω 2 ϕ̇
For this case then some algebra gives the result
# (3x − 2#)
xC (x) = (6.28)
3 (2x − #)
A small table of this function looks as follows
xC (0) = 2#/3, (6.29)
xC (#/3) = #, (6.30)
xC (#/2) = ±∞, (6.31)
so one sees that for 0 ≤ x ≤ #/3 there is a point on the bat that is at rest immediately after
the hit. This is, of course, a point at which it is good to hold one’s hand since then no part of
the impulse from the ball is imparted to the hand. For #/3 < x < #/2 the instantaneous centre
of zero velocity will lie to the right of the rod (# < xC ).
c) The velocities of the points on the bat after the hit are given by the connection formula for
velocities of a rigid body. For the point P on the bar with x-coordinate xP we find the velocity
S S #
vP = vG + PG × ω = ey + (xG − xP ) ex × ϕ̇ ez = − − xP ϕ̇ ey (6.32)
m m 2
in terms of the centre of mass velocity. Some algebra and use of results above gives
S x 1 xP 1
ẏP = 1 + 12 − − (6.33)
m # 2 # 2
6.3. THE COEFFICIENT OF RESTITUTION 97
To get the speed of the point of impact we simply put xP = x and get
2
S x 1
ẏ = 1 + 12 − . (6.34)
m # 2
For 0 ≤ x ≤ # this function clearly takes maximum values at the end points so the answer is
that hits at x = 0 or x = # gives maximum speed of the point of impact. ✷
This set of equations can be used to express the final velocities ẋf in terms of the
initial ones, ẋi , and e. The result can then be inserted into the expression for the
kinetic energy
1 1
T = m1 ẋ21 + m2 ẋ22 (6.39)
2 2
and the change Q = Ti − Tf calculated. It is more instructive, however, to use the
ideas introduced in connection with the two particle problem. There we showed that
the kinetic energy for two particles can be expressed as follows
2
1 1 1 m1 ẋ1 + m2 ẋ2 1 m1 m2
T = mẋ2G + µ(∆ẋ)2 ≡ (m1 +m2 ) + (ẋ2 −ẋ1 )2 . (6.40)
2 2 2 m1 + m2 2 m1 + m2
That is, it can be expressed in terms of the centre of mass velocity ẋG and relative
velocity ∆ẋ as if there were two particles with masses m = m1 + m2 and µ = mm11+m
m2
2
98 CHAPTER 6. IMPACT
respectively, having these velocities. µ is called the reduced mass. It is now trivial to
calculate the energy change since the centre of mass term will be unchanged while the
relative velocity change is directly given by the definition of e. One finds
1
Q = Ti − Tf = (1 − e2 ) µ(∆ẋi )2 . (6.41)
2
From this one draws the conclusion that when e = 1 the energy loss is zero and the
energy is conserved. Such collisions are said to be elastic. Maximal energy loss occurs
at the opposite extreme when e = 0. The relative velocity after the impact is then zero
and the collision is said to be totally inelastic (or plastic).
Example 6.3 Consider a direct central collision between two bodies A and B in which the
coefficient of restitution is e. During the time interval [−δ1 , 0] the two bodies are in contact
and have negative relative velocities (i.e. they approach each other). From time t = 0 to t = δ2
they are in contact but recede from one another (i.e. they have positive relative velocity), see
figure 6.3. The total collision time is thus τ = δ1 + δ2 .
The motion is assumed to take place along an X-axis. The (x-component of the) impulse
on body A, SA , can be split into two parts corresponding to the approaching stage (1) and the
receding stage (2): SA = SA1 + SA2 and similarly for body B. Show that
SA2 SB2
= =e (6.42)
SA1 SB1
i.e. that the ratio of the receding stage impulse to the approaching stage impulse is the same
as the coefficient of restitution.
Solution: According to formula 6.19 we have
SA1 = mA [ẋA (0) − ẋA (−δ1 )] and SA2 = mA [ẋA (δ2 ) − ẋA (0)], (6.43)
SB1 = mB [ẋB (0) − ẋA (−δ1 )] and SB2 = mB [ẋB (δ2 ) − ẋB (0)]. (6.44)
From these two there follows that SA1 = −SB1 and SA2 = −SB2 and thus that
SA2 −SB2 SB2
= = . (6.47)
SA1 −SB1 SB1
There remains to show that this ratio is equal to e.
Since there is no relative velocity at t = 0 both bodies must move with the same velocity.
This velocity must be that of the centre of mass so that ẋA (0) = ẋB (0) = vG . When this is
inserted into equations 6.43 and 6.44, they, together with equation 6.47, give
) )
SA2 mA [ẋA (δ2 ) − vG ] SB2 mB [ẋB (δ2 ) − vG ]
= = = . (6.48)
SA1 mA [vG − ẋA (−δ1 )] SB1 mB [vG − ẋB (−δ1 )]
From this we get
SA2 ẋA (δ2 ) − vG ẋB (δ2 ) − vG
− = = . (6.49)
SA1 ẋA (−δ1 ) − vG ẋB (−δ1 ) − vG
But if a
b = c
d then a−c
b−d = c
d. (Proof: a
b = c
d ⇒ c = ka and d = kb, but then a−c
b−d = a−ka
b−kb =
a(1−k) a
b(1−k) = b, Q.E.D.) Use of this identity gives us
A B
t > δ
2
A B
t = δ2
A B
t = 0
A B
t = -δ
1
A B
t < −δ 1 X
Figure 6.3: This figure shows an inelastic bounce between bodies A and B moving along the
X-axis. Time increases upwards and the time t = 0 is taken to be the instant at which the two
centres of mass of the bodies are closest together. The duration of contact between the bodies
is τ = δ1 + δ2 where δ1 corresponds to a compression of the bodies while δ2 is the time during
which they decompress. The dashed line indicates the motion of the common centre of mass of
the bodies. In example 6.3 it is shown that the coefficient of restitution e can be expressed as
the ratio of the impulses on the bodies during these two time intervals.
100 CHAPTER 6. IMPACT
x
h
Figure 6.4: The figure on the left refers to problem 6.3. The bat rotates around the end with
the handle, with angular velocity ω, when it hits the ball at a distance x from this end.
Figure 6.5: The figure on the right refers to problem 6.4. A football is given a horizontal
impulse S at height h.
6.4 Problems
Problem 6.1 A straight narrow homogeneous rod of mass m = 75 kg and length
# = 180 cm stands vertically on horizontal ground. It is given a horizontal impulse S
at height h above the ground. For what values of h and S will the top end of the rod
hit the ground first in the ensuing motion?
Problem 6.2 A tugboat of mass 800 tonnes is connected to a ship of mass 23000 tonnes
(1 tonne = 103 kg) by a rope of mass 375 kg. The tugboat reaches a speed of 3 knots (i.e.
5.5 km/h) when the rope suddenly becomes taut so water spurts out of it and the ship
starts to move. Since neither boat moves very far while the rope is taut and transfers
force the process can be approximated as an impact. Assume that a reasonable value
for the relevant coefficient of restitution is e = 0.5.
a) Calculate the speed (in knots) of the ship after the impact.
b) Calculate the average force in the rope while it spurts water if this goes on for 1.5 s,
and express the force in terms of an equivalent mass by dividing by the acceleration
due to gravity g = 9.8 kg/s2 .
Problem 6.3 A bat rotates around the end with the handle, with angular velocity ω,
when it hits a ball at a distance x from this end, as shown in figure 6.4. The moment of
inertia of the bat with respect to the axis of rotation (perpendicular to the bat through
the end with the handle) is J and its length is #. The ball is assumed to be at rest
before the hit. Determine the maximal speed of the ball, as a function of x, assuming
that the coefficient of restitution is e.
A b
SA
d
h
d
Y'
B
SB
X
C
y
Y
SC
Figure 6.6: This figure refers to problem 6.5 and shows a door hinged at A and B which hits
a door stop (knob in the floor) at C.
102 CHAPTER 6. IMPACT
α
S
A
B
Figure 6.7: This figure refers to problem 6.6 and shows the two hinged rods and the impulse
S at A.
floor (a door stop) at a distance y from the axis along the lower edge of the door. This
impact gives the door a horizontal impulse SC normal to the plane of the door such
that the door comes to rest.
a) Determine the reaction impulses SA and SB from the hinges on the door at the
impact.
b) Can one choose y so that one of the reaction impulses vanish?
c) Can one choose y so that the magnitudes of the reaction impulses are equal?
Problem 6.6 Two straight narrow homogeneous rods AB and BC are smoothly hinged
at B. The two rods are lying on a smooth horizontal plane and the rod AB is restricted
to slide parallel to itself along a smooth track in the plane. Suddenly an impulse S is
delivered at A making an angle α with the rod, see figure 6.7. Calculate the resulting
reaction impulse in the hinge B.
6.5. HINTS AND ANSWERS 103
Answer 6.2
a) 0.15 knots.
b) The rope will spurt water while the tension in it increases. This period corresponds
to the time interval t ∈ [−δ1 , 0] of example 6.3. One finds that the force corresponds
to the weight of a mass of 80 tonnes.
$
Answer 6.3 If J/m ≤ # then the optimal distance is
x= J/m
$
1 J
and the speed attained is vmax = 2 m ω(1 + e). If J/m > # (is this really possible?)
,ω(1+e)
then x = # and vmax = 1+m,2 /J
.
Answer 6.4
a) This happens at h = r since the friction force from the ground has no time to produce
any effect on the motion.
b) The ratio of remaining kinetic energy to initial is 3/5.
Answer 6.5 The basic equation 6.19 gives (SA + SB + SC ) ex = 0 − (−m 2b ω ex ). Here
−(b/2)ω ex is the centre of mass velocity of the door before the impact. The equation
6.21 gives rA × SA ex + rB × SB ex + rC × SC ex = −Jz ω ez for this problem, since there
are three points at which the door impacts. It is convenient to move the origin to the
intersection of the Z and Y axes of figure 6.6; the position vectors of the impact points
A, B, and C, then become
rA = d ez ,
rB = −d ez ,
h
rC = y ey − ez .
2
The y- and z-components of the angular momentum principle equation then yield
h
d(SA − SB ) − SC = 0,
2
ySC = Jz ω,
SA + SB + SC = mbω/2
obtained from 6.19 above, one can calculate the answers that follow.
2b b
a) The impulse from the door stop is found to be SC = 3 y (m 2 ω). The answer is
1 2b 1hb b
SA = 1− + m ω ,
2 3y 3dy 2
104 CHAPTER 6. IMPACT
and
1 2b 1hb b
SB = 1− − m ω .
2 3y 3dy 2
b) SA cannot be made zero for positive values of y but SB is zero if
b h 0.8 2
y= 2+ = 2+ = 1.24 m
3 d 3 0.75
c) It turns out that SA = SB has no solution but that SA = −SB when [1−(2/3)(b/y)] =
0 i.e. when
2 2
y = b = 0.8 = 0.53 m.
3 3
This is thus a good place to put the door stop since then the two hinges will share the
reaction impulse evenly.
Answer 6.6
The reaction impulse should be SB = (S/5) cos α.
Chapter 7
Open Systems
This chapter discusses the equations of motion for systems which gain or loose matter
through flow through the bounding surface. Such mechanical systems are called open
systems. Rockets, jet-planes and turbines are examples of such systems. The theo-
retical basis for this type of problem is given by the principles of linear and angular
momentum. Mechanical energy, on the other hand, is rarely conserved since internal
non-conservative forces usually are at work in these systems.
where the sum is over the particles inside the system at time t and where vo can be
regarded as the centre of mass velocity of the open system. We use the notation vo
here since this vector is not the time derivative of rG .
To find the equation of motion for the system S we must study the time rate of
change of the momentum po . When we do this we must remember that the particles
in S are not the same all the time. If we assume that in the time ∆t the open system
S has gained ∆N+ particles and lost ∆N− then
N (t + ∆t) = N (t) + ∆N = N (t) + (∆N+ − ∆N− ) (7.4)
is the number of particles of S at time t + ∆t. We can thus write
N (t+∆t)
po (t + ∆t) ≡ mi vi (t + ∆t) (7.5)
i
105
106 CHAPTER 7. OPEN SYSTEMS
Time= t
∆N + + N(t) = Nc
Time= t +∆ t
N(t+ ∆ t) + ∆ N- = Nc
Figure 7.1: This figure illustrates the relationships in equation 7.6. The open system is
represented by the solid line box and moves to the left. The closed system with a fixed number
of particles Nc , corresponding to the time ∆t, is indicated by the dashed line box.
for the momentum of S at time t + ∆t where the sum now is over the particles in S at
this time.
The equation of motion for S follows from the momentum principle in the general
form ṗ = Fe but it must be remembered that the expression p = mvG cannot now
simply be differentiated as was the case for a closed system of particles. To get around
this problem we define a closed system for the time interval t to t + ∆t as the set of
all Nc particles that have been in the system S in this time interval. This number is
clearly
Nc = N (t) + ∆N+ = N (t + ∆t) + ∆N− (7.6)
i.e. the number of particles in the system at t plus those gained at t+∆t, or the number
of particles in the system at time t + ∆t plus those lost since t. This is illustrated in
figure 7.1.
The total momentum of this closed system is by definition
Nc
pc (t, ∆t) = mi vi (t), (7.7)
i
and it is on this system that some total external force Fe may act so that
ṗc = Fe . (7.8)
When the time interval ∆t → 0 the momentum of the closed system we have defined
becomes identical to that of the open system,
but the time derivative of the two momenta are not the same as we’ll now see.
7.1. THE MOMENTUM PRINCIPLE FOR OPEN SYSTEMS 107
Nc
Nc
∆pc = pc (t + ∆t, ∆t) − pc (t, ∆t) = mi vi (t + ∆t) − mi vi (t) = (7.10)
i i
N (t+∆t) ∆N− N (t) ∆N+
mi vi (t + ∆t) + mj vj (t + ∆t) − mi vi (t) + mk vk (t) (7.11)
i j i k
∆N− ∆N+
= po (t + ∆t) + mj vj (t + ∆t) − po (t) + mk vk (t) , (7.12)
j k
where we have used equation 7.6. We have thus found that the changes in the momenta
of the closed and the open systems are related as follows:
∆N− ∆N+
∆pc = ∆po + mj vj (t + ∆t) − mk vk (t). (7.13)
j k
Here ∆po is the change of the momentum of the open system, the first sum is over the
particles that have been lost by the system in the time from t to t + ∆t, and the second
sum is over the particles that have been gained in the same time.
If the system consists of a lot of particles with different velocities this is all one can
say about the change in its momentum. We now assume that we can make a continuum
approximation of the matter that flows in and out of the system. We also assume that
there is only a finite number n− of velocities va− with which the masses ∆m− a flow out
in time ∆t and a finite number n+ of velocities vb with which the masses ∆m+
+
b flow
in. We can then rewrite the above equation in the form
n− n+
∆pc = ∆po + ∆m− −
a va − ∆m+ +
b vb . (7.14)
a=1 b=1
The first term on the right hand side is the momentum change of the open system S
which we can write
∆po = m(t) ∆vo (t) + ∆m(t) vo (t). (7.15)
One should note that the net mass change
n+ n−
∆m = ∆m+
b − ∆m−
a (7.16)
b=1 a=1
contributes to this change in the momentum po but that ∆pc = ∆po even if ∆m = 0
since the velocities of the in and out flows also contribute.
We now divide equation 7.14 by ∆t and take the limit ∆t → 0. If we denote the
mass flows by
∆m− ∆m+
qa− ≡ lim a
and qb+ ≡ lim b
(7.17)
∆t→0 ∆t ∆t→0 ∆t
we get
n− n+
ṗc = ṗo + qa− va− − qb+ vb+ . (7.18)
a=1 b=1
-
v F
ey
v+
ex
Figure 7.2: This figure illustrates the situation in example 7.1. Water flows through the 90
degree bend in the pipe. What is the force needed to keep the pipe at rest? The dashed line
shows the boundary of the open system of pipe with moving water that one must consider.
This is thus the basic equation of motion for the open system S. Note that the mass
flows q are defined to be positive (q ≥ 0) and that all velocities in this equation are
with respect to an inertial system. In the limit ∆t → 0 equations 7.15 and 7.16 give
These together with the equation of motion 7.19 provide sufficient information for
solving many open system problems.
Example 7.1 Water flows at a rate of q = 1 kg/s through a pipe of constant cross-section A
in which there is a 90 degree bend. The speed of the water, which is assumed uniform in the
pipe, is v = 1m/s. Calculate the external force F needed to keep one meter of the pipe around
the bend at rest. See figure 7.2.
Solution: Since in this case the centre of mass velocity vo of the open system is constant,
and since the loss of mass is exactly balanced by the gain its mass m(t) is constant, and we get
that ṗo = m v̇o + ṁ vo = 0. Equation 7.19 now gives
0 = F − qv− + qv+ . (7.22)
With suitably directed basis vectors this gives
0 ex + 0 ey = Fx ex + Fy ey − (−qv) ex + qv ey (7.23)
√ √
and thus Fx = −qv and Fy = −qv. The magnitude of the force is thus F = 2qv = 2N, and
this is the answer. ✷
Equation 7.19 can also be written in an alternative form if we note that ṗo =
m ao + ṁ vo so that
n− n+
mao = F − ṁvo −
e
qa− va− + qb+ vb+ , (7.24)
a=1 b=1
+ −
and that ṁ = q − q . Combining these we get
n− n+
mao = F − e
qa− u−
a + qb+ u+
b , (7.25)
a=1 b=1
where the u− = v− − vo stand for the velocities of the out-flowing matter relative to
the system itself and correspondingly for the u+ .
7.2. ANGULAR MOMENTUM OF OPEN SYSTEMS 109
Example 7.2 A rocket has a motor which ejects mass with a constant rate q with a constant
speed u relative to the rocket. It rises vertically starting with zero velocity at time t = 0.
Calculate its velocity as a function of time.
Solution: Equation 7.25 gives
dvo
m(t) = m(t) g − q u− . (7.26)
dt
With the Z-axis vertically upwards we have vo = vo (t) ez , g = −g ez , and u− = −u ez . The
mass of the rocket is
m(t) = m0 − qt (7.27)
so that q = − dm
dt . All this gives us
dvo dm
m(t) = −m(t) g − u. (7.28)
dt dt
Multiply by dt, divide by m and integrate from t = 0 to t to get
t t
dm
dvo = (−g dt − u ). (7.29)
0 0 m
This gives
m(t) m0
vo (t) = −g t − u ln = −g t + u ln . (7.30)
m0 m0 − qt
This solution is of course only valid until the time t = T at which all the fuel mf = qT is gone.
✷
where, again, the sum is over the particles in the system at time t. We also use the
closed system corresponding to the time interval t to t + ∆t. It will have the angular
momentum
Nc
Lc (t, ∆t) ≡ ri (t) × mi vi (t). (7.32)
i
in the angular momentum case. We assume, as before, that the mass flows through the
system boundary can be considered as continuous with mass per unit time q ± flowing
in/out with a finite number of different velocities v± . We now also assume that these
mass flows take place at well defined positions r± . The above equation then gives
n− n+
∆Lc = ∆Lo + r−
a × qa− ∆t va− − b × qb ∆t vb .
r+ + +
(7.34)
a=1 b=1
110 CHAPTER 7. OPEN SYSTEMS
+
q
r+
+
r ϕ
v
ey
R
ex
Figure 7.3: This figure illustrates the situation in example 7.3. The dashed line is the trajectory
along which the bullets move. They hit the wooden carousel and come to rest in it near the
circumference.
Dividing by ∆t and taking the limit now gives the final result angular
momentum
n− n+
principle for open
L̇o = M − e
r−
a × qa− va− + b × qb vb
r+ + +
(7.35) system
a=1 b=1
Example 7.3 A carousel (merry-go-round) of radius R can rotate freely around a vertical
axis. The moment of inertia with respect to this axis is J0 . Find the expression for the angular
acceleration ϕ̈ of the carousel if it is shot at by a machine gun which produces a horizontal
mass flow q + of bullets with speed v + . Also find the limiting angular velocity if the shooting
goes on for long. The bullets hit the carousel at a perpendicular distance r from its centre, see
figure 7.3, and are embedded in it near the circumference.
Solution: If we assume that the shooting started at time t = 0 the moment of inertia of the
carousel will be
J(t) = J0 + q + tR2 (7.36)
since the bullets increase the mass at radius R. The angular momentum of this open system
(carousel under fire) is
Lo = J(t)ϕ̇(t) ez , (7.37)
with respect to an origin at its centre. Equation 7.35 now gives
d
(J(t)ϕ̇(t) ez ) = r+ × q + v+ = −rq + v + ez . (7.38)
dt
This gives
(J0 + q + tR2 )ϕ̈ = −rq + v + − q + R2 ϕ̇ (7.39)
and this is the desired differential equation for ϕ̈.
If ϕ̇(0) = 0 the angular acceleration is negative for small t > 0. This gives a negative angular
velocity ϕ̇ as the sign conventions of the figure 7.3 demands. However, when the angular velocity
becomes increasingly negative, the right hand side of the equation for ϕ̈ decreases. Eventually
the acceleration will approach zero. When ϕ̈ = 0 the above equation gives the limiting angular
velocity
rv +
ϕ̇lim = − 2 (7.40)
R
and this concludes our example. ✷
7.3. PROBLEMS 111
2r
h v
ω
Figure 7.4: The figure on the left refers to problem 7.2. The water flows out through the hole
of cross sectional area a. The height of the water surface above the hole is h.
Figure 7.5: The figure on the right refers to problem 7.3. A toroidal space station is given an
angular acceleration from zero angular velocity up to a value that corresponds to the acceleration
of gravity on the perimeter.
7.3 Problems
Problem 7.1 A jet plane of mass m is propelled by maintaining an air flow of q mass
units per unit time through its engine. The speed of the air relative to the engine is
w = const.. Calculate the speed v(t) of the plane as a function of time assuming the
initial condition v(0) = v0 . Neglect the effects of fuel consumption and air resistance.
Problem 7.2 A vehicle is propelled by letting water flow out backwards from a tank
of rectangular horizontal cross sectional area A that has collected rain water, see figure
7.4. The cross sectional area of the hole through which the water flows out is a and √ we
assume that a A. One can assume that the water flows out with the speed u = 2gh
(see example 8.1) relative to the tank, where h is the height of the water surface above
the hole. Show that the acceleration of the vehicle must be less than (a/A)2g.
Problem 7.3 A toroidal space station has (outer) radius r, initial mass m, and radius
of gyration d. To achieve an angular velocity ω, corresponding to an artificial gravity
g, the space station, which initially is at rest, is accelerated by means of two identical
rockets on the outer rim. The two rockets are placed at opposite ends of a diameter,
see figure 7.5, and are directed along the tangent of the circumference. Each rocket
has mass flow q (mass per unit time) and exhaust velocity u (relative to the rocket).
Assume that all expelled mass was contained in the rockets (i.e. at radius r). How long
must the rockets be turned on? (Note that rotation with angular velocity ω gives an
artificial ‘gravity’ corresponding to the centripetal acceleration rω 2 .)
Problem 7.4 A chain has constant mass per unit length λ. It is held at one end so
that it hangs vertically with its lower end at a height h above a horizontal table surface.
It is released from rest and starts to fall onto the table. Find the force F (x) from the
chain against the table as a function of the length x of the chain that has reached the
table.
112 CHAPTER 7. OPEN SYSTEMS
Answer 7.2 Denote the density of the water by and the speed of the vehicle by v.
Use of equation 7.25 then gives mv̇ = qu where q = au so that v̇ = au2 /m. The
mass of the vehicle is m = m0 + Ah where m0 is the remaining mass when the height
of the water surface above the hole has become zero. Since u2 = 2gh we find
a2gh
v̇ = .
m0 + Ah
Since m0 ≥ 0 we find that v̇ ≤ a
A 2g and this is what we wanted to show.
Answer 7.3 Use equation 7.35 with Me = 0 and with Lo = J(t)ω ez . One finds that
J(t) = md2 − 2(qt)r2 and thus that the z-component of the equation of motion is
d
[J(t)ω(t)] = −2rq(rω − u).
dt
Some calculations then give
2rqu
ω̇ =
md2 − 2qtr2
and integration yields
u md2
ω(t) = ln .
r md2 − 2qrt2
g
If one now puts ω = r and solves for t one finds that
√
md2 gr
t= 2
1 − exp −
2qr u
is the required burning time of the rockets.
Answer 7.4 Use energy conservation to get the speed of the chain. Consider the
following forces on the heap of chain on the table: the normal force N (x) from the
table, the weight of the heap of chain, gλx, and the the ‘force’ due to the fact that the
heap of chain gains mass. Use of equation 7.25 now gives
for the normal force from the table on the chain heap. The force from the chain heap on
the table F (x) is then of equal magnitude according to the law of action and reaction.
Chapter 8
The mechanics of fluids (gases and liquids) belongs to continuum mechanics, that is,
the number of degrees of freedom can be regarded as infinite. This means that fluids
must be described by fields. The most important kinematic object is the velocity field
v(r, t) which gives the velocity vector at each point r of the fluid at time t. The
mathematical techniques required to handle fluids are those of vector analysis. This
chapter presents derivations of some equations which govern the behavior of the velocity
field. The scalar fields of (mass) density (r, t) and pressure p(r, t) are introduced. The
distinction between laminar and turbulent flow is explained as well as the concepts of
ideal contra viscous fluids.
113
114 CHAPTER 8. THE MECHANICS OF FLUIDS
Let us investigate how this volume changes if we allow the region Ω and its bounding
surface S ≡ ∂Ω to follow the fluid in its motion. Consider a small element of area dA
of the surface S and denote by en the outward directed normal to the surface S at dA.
One then defines the vector element of area
dA ≡ en dA. (8.2)
The change in volume of Ω due to the fluid flow during the time interval ∆t will be
localized at the bounding surface and it is easy to see that, at a given element of area
dA of the surface S, the change in volume will be given by
This volume is the base area dA times the ‘height’, that is, the component of the
displacement dr = ∆t v along the unit normal en . It is negative if the flow has a
component into the region Ω since then v has a negative projection on en . It is zero if
the flow is parallel to the surface, and it is positive if the flow has a component out of
the region. The total change in volume in time ∆t is then the sum of all these changes
in the limit when the dA → 0, i.e. the surface integral
∆V = ∆t v · dA. (8.4)
S
This shows us that the divergence of the velocity field, div v ≡ ∇ · v is a measure of
the change of volume of the fluid.
If the fluid moves like a rigid body we know that its velocity field must have the
form
v(r) = v(0) + ω × r (8.7)
(see equation 2.74). The divergence is then identically zero ∇ · v = 0, which is natural
since the parts of a rigid body have fixed volumes. Another important quantity in
vector calculus if the ‘curl’ (or rotation) of a vector field: curl v ≡ ∇ × v. For the case
of ‘rigid’ flow one finds that
From this one concludes that, in general, the curl of the velocity field, which sometimes
is called the vorticity, is twice the local angular velocity of the fluid. Fluid flow for
which ∇ × v = 0 is said to be ‘irrotational’.
the readings of a thermometer at some fixed point of space corresponds to the partial
derivative
∆T (r, t) ∂T
lim = (8.9)
∆t→0 ∆t r=const ∂t
of the scalar field T (r, t) with respect to time.
If, on the other hand, one measures some quantity, such as temperature for example,
in some (material) fluid element with trajectory r(t) and considers the time rate of
change of these measurements one finds what is called the total time derivative
∆T (r(t), t) dT (r(t), t)
lim = = (8.10)
∆t→0 ∆t dt
∂T dx ∂T dy ∂T dz ∂T ∂T
= + + + = + v · ∇T. (8.11)
∂t dt ∂x dt ∂y dt ∂z ∂t
This quantity is sometimes referred to as the ‘substantial’ or ‘material’ time derivative.
This time rate of change of temperature is measured by a small light thermometer
which follows the fluid in its flow. We will write total time
derivative
d ∂
≡ +v·∇ (8.12)
dt ∂t
for the total time derivative operator.
where the minus sign is due to the convention that the area element vector dA is
directed outwards, so that v · dA is positive where matter flows out. Equating these
two expressions and using Gauss’ theorem in the form Gauss’ theorem
v · dA = ∇ · (v) dV, (8.16)
S Ω
gives us
∂
+ ∇ · (v) dV = 0. (8.17)
Ω ∂t
Since the region Ω is arbitrary one finds that mass conservation is expressed by the
partial differential equation equation of
∂ continuity
+ ∇ · (v) = 0 (8.18)
∂t
relating the partial derivatives of the fields and v. This equation is called the equation
of continuity.
116 CHAPTER 8. THE MECHANICS OF FLUIDS
Here we follow a given fluid particle with constant mass, dm = dV =const., so the
time derivative is given by
d ∂
dṗ = [( dV )v] = ( dV ) + v · ∇ v, (8.20)
dt ∂t
where we have used equation 8.12 for the total time derivative. This is thus the time
rate of change of the momentum for the fluid particle in dV .
dW = ( dV ) g, (8.21)
where g is the acceleration due to gravity. In continuum mechanics this type of force is
called a volume (or body) force since it is proportional to the volume dV . Apart from
8.2. THE EQUATION OF MOTION FOR AN IDEAL FLUID 117
body forces there are also surface forces acting on the elements of the fluid. These are
of two basic types: those perpendicular to the surface, that is, parallel to the vector
surface element dA, and those tangent to the surface. The first type is due to the
pressure p(r, t) and the total force on the fluid in the region Ω due to pressure is force due to
pressure
Fp = −p(r, t) dA (8.22)
S
where S is the surface surrounding Ω. The minus sign here is needed because of the
convention that the vector surface element points out from the region and we want the
force on the fluid inside S from the rest of the fluid. Let a be a constant vector. Then
use of Gauss’ theorem gives us
a· p dA = (pa) · dA = ∇ · (pa) dV = (8.23)
S S Ω
= a · ∇p dV = a · ∇p dV. (8.24)
Ω Ω
The surface force due to pressure on the small element of fluid in dV can thus be written
1 2
∇v = v × (∇ × v) + (v · ∇)v, (8.28)
2
where v = |v| and this allows us to rewrite Euler’s equation above in the form
∂v 1 2 ∇p
+ ∇v − v × (∇ × v) = g − . (8.29)
∂t 2
We will now investigate one of the special cases in which this equation has a first
integral.
118 CHAPTER 8. THE MECHANICS OF FLUIDS
Z
A
O
h
u
B
Figure 8.1: This figure illustrates the situation in example 8.1. There is a small hole in the
tank at depth h below the liquid surface.